先验知识在遥感反演中起到重要的作用，本文首先根据各向异性平整指数AFX和地表多角度观测数据集建立了BRDF二向性反射分布函数原型库，然后介绍了以BRDF原型作为先验知识的提取多角度数据地表反照率的原型反演算法，为验证算法的精度，将原型反演及全反演算法拟合验证数据子集得到的反照率与MODIS全反演算法拟合完整数据集得到的结果作比较。结果对比表明，观测天顶角在40°；以内时，非垂直主平面的观测数据能够代表部分地表的各向异性反射特征，此时原型反演算法的最大平均绝对误差为0.036，相对精度比全反演算法高约5%—10%；当数据位于垂直主平面时，观测信息量严重不足，此时原型反演算法结果相对较稳定，而全反演算法的最大绝对误差高达0.18。 综上所述，原型反演算法能够在整个采样空间范围内对BRDF进行约束，对观测数据位置分布的依赖较小，抗噪声能力强，在反演观测信息不足的多角度数据时，结果精度优于全反演算法。

A priori knowledge plays an important role in the retrieval of surface albedo from satellite observations. Here a BRDF archetype database is built according to Anisotropic Flat Index (AFX) which can identify surface reflectance anisotropy. Then a method to derive land surface albedo from spare observations and BRDF archetype is presented. Albedos from all the observations and full inversion method are used to validate the albedo estimation from our method. Compared results show that, when the observations that have view zenith angles less than 40° are not on the cross-principle plane can only represent part of reflectance anisotropy. Compared with full inversion method, the maximum absolute error of BRDF archetype inversion method is 0.036, and the relative error is improved about 5%—10%. For those observations are located on the cross-principle plane, the largest absolute difference of full inversion method is up to 0.18 due to the insufficient and spare angular samples. However, the BRDF archetype inversion method appears more accurate. In sum, BRDF archetype inversion method constrains BRDF shape over the entire space and was less dependent on the distribution of the observations. It was also less sensitive to noise. For the multi-angle data that have insufficient information, BRDF archetype method is preferred than full inversion method.