This paper presents new ideas and recent progress in our joint effort to model the interaction mechanism of radiant flux with discontinuous canopies. At the scale of a small volume in which leaves can be regraded as homogeneously distributed

some recently develop-ped RT （Radiative Transfer） theories are the btst to model the phenomenon by using descriptors such as leaf scattering characteristics

leaf size

volume density of leaf area

etc. However

at the scale of a stand or TM pixel

GO （Geometric Optical） models catch the basic features of such stands under sunlight

i.e.

foliage are clustered into crowns and crowns cast shadows. Hence in present practi ce

simple pure GO models are the only models which can be applied for natural discontinuous canopies.However

pure GO models require signatures of sunlit and shaded crown surface and background as known parameters under given situation or to be determined in situ. This has been proven a major restricting factor in applications and model inversion. And it is also a drawback that these signatures have not been related to leaf descriptors

wavelength

and sky conditions.In this paper

we proposed an integrated GO-RT approach to model the radiation climate in a discontinuous canopy. A key link in this approcah is gap probability （Pgap） model which we developped earlier. Gap probability

on one hand

can be obtained through pure GO model which reflects the structure at the stand scale. On the other hand

gap probability within crown is closely related to the pro cess that radiation collides and is scattered by foliage at the scale of RT models. Hence it becomes a natural link between two kinds of models best at the corresponding scales.In Li and Straheler （1988）

Pgap was modeled at the ground only. In this paper

the way to model Pgap at any height within canopies is developed. Vertical distributions of sunlit crown surface and within-crown pathlength distributions for different canopy structures are demonstrated. Since the penetration depth of direct sunlight into canopies play an important role in many fields other than remote sensing

we present here the results at this early stage of our modeling work to share with our colleagues in China for comments and possible validations. The way to model multiple scattering is also suggested in this paper. The concept of "Qpenness（Kopen） distribution" is introduced

and the （Kopen） at ground has been applied to shrub sites in HAPEX-Ⅱ-Sahel experiment and got good results. （Franklin et al. submitted to Agricultural and Forest Meteor.

Feb.

1993）. The approach is still under development

though the initial results are encouraging.