The leaf area index (LAI), defined as half the total developed area of green photosynthetically active elements per unit horizontal ground area, is one of the key biophysical variables of vegetated surfaces. Optical devices developed to overcome the burden of time consuming, expensive and difficult to conduct sampling in tree canopies are based on the unrealistic assumption that leaves are uniformly distributed in the canopy. This assumption is violated when the leaf area density varies in the horizontal plane due to the clustering of leaves in trees. In this work, a hierarchical model in which leaves are represented as a second level Boolean model whose centers are distributed conditional on a first level Boolean model representing crowns is proposed. Crowns will be furthermore modeled as concentric rings with varying leaf density. Analytical expressions relating second order functions, such as variograms or covariance functions, to canopy structure characteristics such as LAI, leaf size, crown cover and crown radius will be established. From the fitting of these second order functions, the proposed Boolean model will be inverted to retrieve the LAI and canopy structure characteristics. The methodology is assessed over a number of simulated test cases including realistic 3D canopy structure of forest canopies.