The geometric characterization of tree crops is a necessary task to obtain simple though valuable variables such as height and width of tree tops and more complex variables such as porosity and tree top surface, in which this paper focuses. To obtain these variables, a ground-based mechatronic LiDAR system and a multi-size voxels algorithm have been developed. We test the LIDAR system and our algorithms for two cases: a single tree case and a multi-tree case. In the latter, we apply our algorithms in a tree row from an avocado grove in Chile, showing that our system is portable, accurate and can offer the farmer a useful tool for crop monitoring. In addition, we compare our approach with others previously published, showing that our system is more efficient when estimating porosity and crown surface, and offers more capabilities for the decision making process in agricultural activities.