A four-level bidirectional reflectance model based on canopy architecture and its inversion

Open boreal forests present a challenge in understanding remote sensing signals acquired under various solar and view geometry. Much research is needed to improve the ability to model the bidirectional reflectance distribution for retrieving the surface information using measurements at a few angles. The geometric-optical bidirectional reflectance model outlined in this paper differs from Li-Strahler's model in the following respects: (i) the assumption of random spatial distribution of trees is replaced by the Neyman distribution which is able to generate the clumpiness of a forest stand; (ii) the multiple mutual shadowing effect among tree crowns is considered using the combination of the negative binomial and the Neyman distribution theory; (iii) the probability of observing the sunlit background is modelled using a canopy gap size distribution function which is shown to affect the size and width of the hotspot; (iv) the branch architecture of conifer trees affecting the directional reflectance is simulated using a simple angular radiation penetration function; (v) the tree crown surface is treated as a complex surface with small structures which themselves generate mutual shadows and a hotspot. All these levels of canopy architecture are shown to have important effects on the directional distribution of the reflected radiance from boreal forests. The model results compare well with a data set from a spruce forest. The model after validation is used as a tool to retrieve the leaf area index (LAI) of plant canopies according to a vegetation index calculated from the modelled red and near infrared bidirectional reflectance factors at various solar and view angles. The importance of canopy architecture (tree density and distribution, and branch and shoot structure) in the retrieval of LAI will be demonstrated.