Actually this is not what I research, but what caugth my attention were the Figure 5. I wonder if there is a way to group the same frequencies using clusterization, or using Principal Component Analysis (PCA) to get clear and condensate data. Maybe PCA could give you the more relevant results from these you alrady have. Good luck.
Hi Fabiane, Challenging problem! Google (concrete spectral reflectance) came up with some info and graphics that were all new to me. While likely way off topic (maybe related to one of your other projects!), found this recent lidar / road paper quite interesting.
Best,
... Mike
Article Forest road quality control using ALS data