双站TanDEM-X InSAR系统已成功应用于生产全球数字高程模型。然而，受X波段SAR信号的穿透能力限制和森林体积散射的影响，在森林地区提取的DEM包含严重森林信号。因此，利用TanDEM-X InSAR数据估计林下地形的关键在于如何降低森林体散射对InSAR测高的影响。鉴于此，本文提出了一种基于机器学习联合TanDEM-X InSAR、ICESat-2和Landsat 8数据估计林下地形的方法。为验证所提方法的有效性，选用了两个具有不同地形条件和森林类型特征的试验区进行了测试，并利用高精度机载LiDAR DTM进行精度评定。在加蓬热带雨林试验区，估计的林下地形RMSE为5.45 m和5.91 m，与InSAR DEM (14.70 m和18.58 m) 相比提高了60%以上；在西班牙北方森林试验区，林下地形的RMSE也从6.05~9.10 m降低到了3.06~4.42 m。

Digital elevation models (DEMs) are indispensable data source for natural resource investigation, climate change analysis, disaster monitoring and assessment. TanDEM-X, as the first twin-satellite synthetic aperture radar interferometric (InSAR) system, has been successfully obtained a high-precision global DEM with 12 m resolution. However, the penetrability of short-wave signals is limited, and the DEMs obtained in dense forest areas are typical of low precision, contaminated by the forest canopy signal. The phase center height (PCH) height caused due to the forest volume scattering needs to be removed from InSAR-derived DEM to obtain sub-canopy topography. Unfortunately, the TanDEM-X global standard mode acquires single-baseline, single-polarization data, which is difficult to satisfy the existing model calculations, and external data needs to be introduced. In this paper, we propose a machine learning-based method to estimate sub-canopy topography by combining TanDEM-X InSAR data, ICESat-2 data, and Landsat 8 OLI data. The effectiveness of the proposed method is tested and validated in two test sites with different topography properties and forest types. In the Estuaire rainforest test site, compared with the airborne LiDAR digital terrain model (DTM), the root-mean-square errors (RMSEs) of the InSAR DEMs corresponding to two locations are 14.70 m and 18.58 m. After removing the PCH, the accuracy is improved to 5.54 m and 5.86 m, which represents an improvement of over 60%. In the Spanish northern forest test site with complex terrain, the RMSE of sub-canopy topography also increased from 6.05~9.10 m to 3.06~4.42 m. In addition, we investigated the necessity of the proposed method to use InSAR observations and the impact of the accuracy of the ICESat-2 control points used on the sub-canopy topography estimation. These satisfactory results demonstrate the potential of the proposed method to estimate sub-canopy topography for future spaceborne InSAR missions (e.g. TanDEM-L and LT-1), when only single polarization and baseline data are available. Additionally, combining the high resolution of the TanDEM-X with the strong penetrability of the BIOMASS, the sub-canopy topography with higher accuracy and resolution will be estimated in the future.