A botanical plant modelling system for remote sensing simulation studies.

摘要

A modelling system (the Botanical Plant Modelling System, BPMS) is described, which is designed to simulate the radiation regime of plant canopies using three-dimensional models of the geometric structure of individual plants and Monte Carlo ray tracing techniques. The aim is to provide a tool in which no (major) a priori assumptions are made about the nature of the radiation regime, to be used with explicit (deterministic) descriptions of plant and substrate geometric and radiometric properties. The BPMS can be used to investigate the behavior of canopy reflectance for any remote sensing scenario, as well as to test the validity and applicability of assumptions made in simpler models. The modelling works by: (i) forming a description language to represent plant topology and geometric attributes, in which any element of the plant can have any specified radiometric characteristics; (ii) modelling individual plant geometry using three-dimensional representations of plant leaves, stems etc. derived from this description; (iii) distributing these individual plants within a canopy according to any desired planting pattern over any terrain; (iv) producing radiometric simulations according to given sensor characteristics and illumination conditions. The model is used to investigate the nature of the radiation regime in two cases: (i) a simple surface composed of opaque spheres; (ii) a canopy of sugarbeet plants, in an analysis of the variations in directional reflectance due to row-planting and leaf wilting effects. The results demonstrate that the multiple-scattered component of radiation in the former cannot easily be modelled using standard analytical formulae. In the latter, it is demonstrated that the magnitude of row effects is reduced with increasing leaf wilt, due to an increase in the proportion of ground cover. Simple methods such as soil-adjusted vegetation indices go some way in dampening the effects, but an analysis of the vegetation amount is not straightforward without the use of an explicit model which deals with variations in the proportion of sunlit and shaded components viewed. The BPMS is shown to be very well-suited to experimentation of this sort.