In order to classify tree types, we introduce a new technique, which detects the change of Polarimetric Entropy parameters. We define the difference of the two-dimensional polarimetric entropy parameters, and applied the proposed technique to Pi-SAR airborne polarimetric SAR data. The result in an urban area shows that the seasonal change is very random and we could not find significant changes between two scenes. Then, we applied the technique in a forested region. The change in coniferous trees and broadleaf trees is statistically different. Using the change in polarimetric entropy properties, we find this technique can be used for tree type classification.