Abstract: This article uses the 2020 L-band fully polarized ALOS-2 of the Saihanba Mechanical Forest Farm in Hebei Province as the data source, and mainly classify larch forests, broad-leaved forests, mixed coniferous forests, etc., compares the classification accuracy of 7 kinds of ground objects by three schemes of multi-temporal polarization decomposition combined with interference coherence, single-temporal polarization decomposition and single-temporal polarization decomposition combined with interference coherence. The methods are as followed:（1） calculate 12 polarization decomposition methods（81 polarization decomposition features） for each time-phase full polarization data;（2） combining the backscattering coefficient and interference coherence with the selected polarization decomposition features respectively, three classification schemes are obtained, and the features involved in the classification are screened according to their importance combined with the random forest classifier;（3） determining the best classification scheme for land cover types within the study area. The results show that: under temporal stage, according to the importance of the selected features, the features participating in the classification are increased, and the overall classification accuracy first increases and then stabilizes. The temporal information can improve the polarization decomposition classification accuracy of polarimetric SAR images. The overall classification accuracy and Kappa coefficient increase from 87.33% and 0.851 0 to 90.41% and 0.887 5, respectively. Combined with random forest classifier for feature screening, 102 classification features in 5 phases are obtained, and the best classification accuracy is obtained（the overall classification accuracy is 93.84%, and the Kappa coefficient is 0.927 6）. Multi-temporal interferometric coherence combined with multi-temporal polarization decomposition can effectively improve the identification accuracy of land cover types and can also provide a reference for the rapid identification of stand types.