单坡面冠层二向性反射模型是山区植被遥感反演的基础，对于了解山区地表反射特性具有重要作用。依据建模原理的不同，现有的单坡面冠层二向性反射模型主要可分为辐射传输、几何光学和混合模型三类。第一类模型主要适用于均匀的连续冠层场景，在多次散射刻画过程中具有优势。第二类模型将树冠视作规则几何形体，仅适用于非连续冠层场景，在单次散射刻画过程中具有优势。第三类模型耦合了辐射传输与几何光学模型，但依然受到冠层场景假设的限制，不能既适用于坡面连续冠层场景又适用于坡面非连续冠层场景的模拟。随机辐射传输模型（Stochastic Radicative Transfer Model, SRT）具备适用上述两类冠层场景，可考虑天空光入射贡献，多次散射刻画准确的优势，因此在SRT基础上，建立适用于坡面的坡面随机辐射传输模型（Topographic Stochastic Radiative Transfer Model, TopSRT）能克服已有单坡面冠层二向性反射模型的缺陷。 计算机真实结构模拟基于严密的物理推导，经过了广泛的验证，能准确刻画三维场景真实结构及辐射特性，在山区仅需通过高性能计算即可获得多类场景的模拟结果，具有较高的可信度。在山区基于计算机真实结构模拟来建立参数化的物理模型具有广阔的应用前景，代表着山地遥感建模的未来发展趋势。本文利用高性能的基于光线追踪原理的计算机模拟软件DART（Discrete Anisotropic Radiative Transfer）和LESS(Large-scale Emulation System for forest Simulation)分别模拟基于真实结构的连续和非连续冠层场景，将其结果与两大单坡面几何光学模型（GOMS、GOST）对比，发现GOMS模型热点较宽泛，GOST模型模拟耗时长，模拟结果不够稳定等问题，进一步经过以下四步将适用于平地的SRT模型拓展到了坡地： （1）平面坐标系向坡面坐标系的角度转换；（2）条件对相关函数和分层覆盖度的修正；（3）叶倾角分布函数和热点因子修正；（4）坡面二向性反射因子（BRF）与平地BRF的转换。首先假设树冠叶倾角为球形分布，不考虑天空光，分别对比了计算机模拟结果与建立的TopSRT在连续和非连续冠层场景的模拟结果，发现TopSRT与计算机模拟模型高度一致，红光波段的RMSE小于0.007，近红外波段RMSE小于0.07，相关系数约为0.9。其后假设树冠形状为圆柱体，验证了平地和坡地的不同天空光比例场景下，TopSRT与计算机模型模拟结果随天空光入射贡献变化的一致性。最后分析了TopSRT模型随地形、树冠结构参数的变化，与理论预期一致。TopSRT广泛适用于连续和非连续场景的单坡面二向性反射模拟，维持了树冠向地性生长的生理特性，还可考虑天空光入射贡献，计算效率高，克服了已有单坡面模型受模拟场景限制的缺陷。

Solo slope surface bidirectional reflectance factor (BRF) model is the basis of remote sensing inversion of vegetation in mountainous areas, and plays an important role in understanding the surface reflection characteristics of mountainous areas. According to the different modeling principles, the existing solo slope surface BRF model can be divided into three categories: Radiative Transfer Model (RT), Geometric-Optical Model (GO), and Hybrid Model. The first kind of model is mainly suitable for uniform continuous canopy scenes, and has advantages in the process of multiple scattering characterization. The second kind of model regards the tree crown as a regular geometric shape, which is only suitable for discontinuous canopy scenes and has advantages in single scattering characterization. The third kind of model is a composition of radiation transfer and geometric optics model, but it is still limited by the assumption of canopy scenes. It cannot be applied to both continuous and discontinuous canopy scenes on slopes. However, the Stochastic Radicative Transfer Model (SRT) has the advantages of being suitable for the above two types of canopy scenarios, considering the contribution of sky light and accurately depicted multiple scattering. Therefore, based on the model theory, Topographic Stochastic Radiative Transfer Model (TopSRT) can overcome the shortcomings of the existing solo slope canopy reflectance model. At present, there are three main verification methods for solo slope canopy bidirectional reflectance model, which are compared with satellite data, ground measurement data and computer simulation. Computer realistic simulation is based on rigorous physical derivation and has been widely verified. It can accurately depict the real structure and radiation characteristics of three-dimensional scenes. In mountainous areas, simulation results of many kinds of scenes can be obtained through high-performance computing with high credibility. Building parameterized physical models based on computer realistic simulation in mountain areas has broad application prospects and represents the future development trend of remote sensing modeling in mountain areas. It is found that the hot spots of GOMS model are broad, the simulation time of GOST model is long, and the simulation results are not stable enough. The SRT model suitable for flat land is extended to sloping land through the following four steps: (1) Transfer plane coordinate system to slope coordinate system; (2) correction of pair correlation function and layered fractional vegetation coverage; (3) correction of leaf inclination distribution function and hotspot factor; (4) conversion of slope BRF to flat BRF. Assuming that the leaf angle distribution of the canopy leaves is spherical and the sky light is absent, the computer simulation results are compared with those of TopSRT in continuous and discontinuous canopy scenarios. It is found that TopSRT is highly consistent with the computer simulation results. The RMSE in red light band is less than 0.007, the RMSE in near infrared band is less than 0.07, and the correlation coefficient is about 0.9. Finally, the TopSRT model changes with topography and canopy structure parameters are analyzed, which is consistent with the theoretical expectations. TopSRT is widely used to simulate the bi-directional reflection of solo slope in both continuous and discontinuous scenarios. It maintains the physiological characteristics of perpendicular growth for the tree crowns. It also considers the contribution of sky light incidence and has high computational efficiency. It overcomes the limitation of the existing solo slope canopy BRF models which are limited by the rigid assumptions.