混合相云的反演一般利用冰云和水云的折射率虚部有不同的谱分布。由于影响因素众多，当前针对混合相云的反演算法有比较大的误差。本研究使用主成分分析对卫星观测数据进行分析得到混合相云的混合比变化，主成分分析得到的主成分系数对混合比变化非常敏感，使用混合比变化作为约束并且使用Levenberg-Marquardt算法反演得到混合相云的微物理参数。. 反演云的混合比需要快速的前项辐射模型。基于小角度近似的辐射传输方法在保持精度的基础上，计算时间得到了极大的提高。目前的方法只是针对单层云，本研究通过把单层云的辐射模型发展到多层云以及与地表的相互作用，为混合相比的反演提供技术基础。

Retrieval of mixed-phase clouds usually use the spectral differences of the imaginary parts of the ice and water clouds. The difference can be used to retrieve mixed-phase clouds. However, current retrieval methods have a lot of errors because a lot of factors influence the retrieval. This study uses the principal component analysis to get the change of mixed ratios of mixed-phase clouds. Using the mixed ratio changes for different samples, we use Levenberg-Marquardt algorithm to retrieve the microphysical parameters of mixed-phase clouds.. Forward radiative transfer model is also necessary to retrieve the cloud. The radiative transfer solution based on small-angle approximations can significantly reduce the computational time with high accuracy. Current algorithm is only for one layer. This study develops this algorithm to multiple layers with the interaction with the surface. This algorithm provides a technique support the retrieval of mixed ratios.