I've noticed an interesting conundrum in my lab regarding GLCM textures from C-band SAR data for LULC classification and change-detection on tropical rainforests. On one hand, it looks it is a must to include these features in the classification tests, however nobody is surprised when they do not improve the results. Do we have enough evidence in literature to support either the need to test it or studies in which they did not improve the results? Can you suggest strong references that used this type os features for LULC classification/change detection on tropical forests?