Objectives Interferometric synthetic aperture radar (InSAR) and optical photogrammetry are two primary techniques for generating digital surface model (DSM). However, InSAR is susceptible to geometric distortion, while optical photogrammetry is limited by cloudy and rainy weather. The fusion of multi-source DSM can effectively leverage the complementary advantages of the data to enhance the quality of DSM derived from a single data source. Traditional weighted average fusion methods rely on prior information for weight allocation and don't adequately consider the impact of terrain features on elevation errors. There is a need to develop a new fusion method.

Methods This paper proposes a stepwise regression fusion method for InSAR and optical DSM that takes topographic features into account. Taking InSAR and optical DSM as research subjects, and considering the commonalities and differences in error characteristics between InSAR and optical DSM, terrain geometric features, land cover features, and elevation difference features are extracted as training samples in the training area. Subsequently, the stepwise regression method is employed to construct a regression model between elevation errors and terrain features by incrementally adding or removing variables. Elevation errors are then predicted to adaptively determine the fusion weights. Finally, fusion experiments are conducted using TanDEM-X DSM and AW3D30 DSM data from mountainous areas in the United States, and the results are tested in two different experimental regions.

Results The proposed method outperforms the traditional weighted averaging fusion method. The root mean square errors decrease by 14.3% and 18% in the two test regions, respectively. Compared with the random forest method, the proposed method achieves higher fusion precision while maintaining favorable interpretability.

Conclusions The proposed method effectively improves the quality of fused DSM by utilizing terrain features and achieves favorable performance in complex terrain areas, providing an effective solution for multi-source DSM fusion.