The forest vertical structure heterogeneity （FVSH） model constructs a forest vertical structure function using tomographic spectra obtained from tomographic synthetic aperture radar（TomoSAR），and can accurately describe the forestvertical structure.The accuracy of polarimetric interferometric synthetic aperture radar （PolInSAR） forest height inversion is closely related to the coherent scattering model.Optimizing the forest vertical structure function with tomographic spectra can improve the inversion accuracy.In this paper，the heterogeneous coherent scattering models constructed from different vertical structure tomography spectra are comparatively analyzed，and the optimized forest vertical structure function and the forest height inversion results of the coherent scattering modelare analyzed.Tests on an airborne P-band fully polarizedsynthetic aperture radar （SAR） dataset covering tropical rainforests in French Guiana are carried out.The results show that themultiple signal classification （MUSIC） algorithm outperforms the Beamforming and Capon algorithms in optimizing forest vertical structure functions and retrieving forest heights based on coherent scattering models.This study is of great significance for the forestry application of domestically produced P-band SAR satellites.