Modeling polarized reflectance of snow and ice surface using POLDER measurements

Abstract A new semi-empirical bidirectional polarization distribution function (BPDF) model for snow and ice surface is proposed in this short communication paper. The new model adopts two free parameters: one for the overall amplitude and the other for the shadowing process. It is special because a new snow index (NSI), which is defined as the simple algebraic ratio between the reflectance at green band and two NIR bands, is added to the new model to compensate for the influence of the snow grain size. The new model is compared with four existing semi-empirical BPDF models (i.e., Nadal-Breon model, Maignan model, Litvinov model and Xie-Cheng model) using the BRDF-BPDF database generated from POLDER measurements. Experiments show that the new BPDF model provides the best performance among these models in terms of the root-mean-square-error obtained for two cases, that is, for data collected at 10 sites and for data collected at all sites. Error distributions under different solar zenith angles also show that the new BPDF model achieves smaller errors in most directions. The new BPDF model is thus a helpful tool for the parameterization of the optical properties of the snow and ice surface.