Abstract: Objectives: LuTan-1 (LT-1) is the world's first L-band bistatic synthetic aperture radar interferometric (InSAR) system, designed for high-precision terrain mapping. However, due to the influence of forest volume scattering, the digital elevation model (DEM) obtained in the forest area contains severe forest height signals and cannot represent the real surface elevation. In addition, the LT-1 InSAR system acquired single-baseline, single-polarization InSAR data in stripe mode 1, which lacks observation information and cannot directly support the solution of existing physical models (e.g. random volume over ground, RVoG). Methods: Given this, this paper used time-frequency analysis to increase observation information or simplify the physical model to estimate sub-canopy topography from LT-1 InSAR data. Specifically, the RVoG model based on sub-aperture decomposition, the simplified C-SINC model, the rational function model and the machine learning model are adopted. Based on the above four models, we quantitatively evaluated the performance of the LT-1 InSAR system in retrieving sub-canopy topography. First, the penetration capability of LT-1 in forested areas and the accuracy of its estimated ground elevation were quantitatively assessed. Next, the performance of the four proposed methods for estimating sub-canopy topography using LT-1 InSAR data was tested and validated at two test sites in Spain with differing terrain and forest conditions. Results: Compared with airborne LiDAR-derived terrain products, the estimated sub-canopy topography accuracy reached 1.75 m and 2.63 m in forested areas, representing an improvement of over 50% compared to the original InSAR DEM. The machine learning model showed the best accuracy in the two test sites, with an accuracy improvement of approximately 54% compared to InSAR DEM. Conclusions: LT-1 exhibits excellent interferometric quality and strong sensitivity to forest physical properties in forested areas, demonstrating great potential for large-scale and high-precision sub-canopy topography estimation. We systematically analyzed the advantages, limitations, and applicable scenarios of different methods, providing valuable insights for the operational production of LT-1 sub-canopy topography products. In addition, the methods proposed in this paper provide methodological support for future domestic and foreign low-frequency SAR satellites (such as Germany's TanDEM-L mission, ESA's BIOMASS mission, and China's P-SAR mission) to estimate high-precision sub-canopy topography.