A forest canopy height surface model for scene simulation

Abstract

A model of forest canopy heights can support realistic remote sensing scene simulation. The final pixel (x, y) array of canopy heights results from merging structural and statistical sub-models. The structural model considers tree crowns to be structural primitives and permits explicit specification of forest structural characteristics (crown forms and spacing). The statistical model allows the inclusion of smaller scale effects (clumping or rough ness), and nonstationary effects can be introduced by merging the structural and statistical model results with spatially varying weights. The model supports several tasks in realistic scene simu lation (e.g., for use with algorithms which use surface geometry to determine pixel brightness). It also helps define realistic forest- top geometry classes or types so that simulations do not require high-level user expertise or detailed specification. The model also provides controlled test-bed data for canopy measurement and analysis technique development. By retaining tree shape and spacing information in a height surface, the model partially fills the gap between stochastic texture modeling and three- dimensional tree (object) models in scene synthesis.