冰雪和海洋表面的二向反射和反照率实用模型研究

Bi-directional reflectance (BRDF) is a common phenomenon of the Earth surface. The BRDF parameter product derived from remote sensing can be used in estimation of many other physical parameters, e.g., albedo, LAI and aerosol optical thickness, enhancing the quality of corresponding data products. The BRDF parameter product released by NASA has achieved great success. However, its algorithm applies the kernel-driven BRDF model to the whole globe while the model is actually developed for vegetation/soil-composed land surface. That means, the model could be invalid in snow/ice, ocean or mixed pixels. And there is no ocean surface albedo product till now. So, there is a space for improvement in the current algorithm for BRDF parameter product. The proposed research aims to develop accurate and invertible BRDF model for snow/ice and ocean, and corresponding inversion algorithms which are potential for operational product generation. Firstly the physical BRDF model for snow/ice and ocean surface should be developed by integrating the geo-optical submodel into the traditional snow/ice model to describe the surface scattering, and the white-cap spectral should be improved in the ocean BRDF model. Secondly, semi-empirical model, which is more practical in operational inversion, should be derived based on the physical model, and the experience in land surface BRDF modeling can be adopted when building the semi-emipical model. Finally, a dynamic inversion algorithm, which is necessary to describe the temporal variation of snow/ice and ocean albedo, should be built with the help of multi-sensor remote sensing data as well as a priori information from historical data and climate simulation. The output of this research will considerably enrich the models and algorithms in quantitative remote sensing. It will also support the scientific community to generate BRDF/albedo parameter product covering the whole globe.

二向反射（BRDF）是地表的基本光学特性， BRDF产品可以支持反照率、叶面积指数、气溶胶光学厚度等其他遥感参数产品的精度提高。目前的BRDF/反照率产品已经取得很大成功，但是其算法只使用了针对陆面的模型，在冰雪、海洋以及混合像元地表上还有待改进，海洋表面尚没有BRDF/反照率的遥感产品。本课题拟发展冰雪、海洋的可反演的BRDF模型：首先完善现有物理模型，在冰雪模型中补充描述冰雪的表面阴影和镜面反射效应的几何光学模型，在海洋模型中改进白帽反射率的参数化方法；然后参照陆面遥感中的线性核驱动模型的思想推导适用于冰雪、海洋BRDF的半经验模型；最后通过联合使用多传感器数据、引入先验知识和气象参数等方式提高反演的时间分辨率，建立能够反映冰雪、海洋表面高度动态特点的BRDF/反照率反演算法。本研究将丰富定量遥感模型与方法，为将来生产完整覆盖全球表面的BRDF/反照率产品打下基础。

二向反射（BRDF）/反照率是地表的基本光学特性，目前的BRDF/反照率遥感产品已经取得很大成功，但是其算法只使用了针对陆面的模型，在冰雪、海洋以及混合像元地表上还有待改进，海洋表面尚没有BRDF/反照率的遥感产品。本项目研究发展了针对冰雪、海洋的可反演的BRDF/反照率模型。并最终基于直接反演算法反演了MODIS遥感数据，生成的全球反照率产品在冰雪、海洋表面具有较高精度，能够反映广阔海洋和高纬度区域反照率的空间分布和时间变化。. 对于海水表面，基于对海水反射光线的物理过程的分解，将海水反射表示为海面耀斑、海面白帽、水体离水反射三个部分的加权和，由海面风场、水色悬浮物以及太阳角度驱动，分别采用文献中最合适的公式进行参数化，计算海水BRDF和反照率，称之为TCOWA模型。对于纯冰雪表面，首先通过输入一系列冰雪物理参数(雪物理参数包括雪粒径、雪密度和污染物含量等；冰物理参数包括盐胞半径、体积百分比、气泡半径、体积百分比和污染物含量等)，采用参数化方法快速计算冰雪的固有光学参数，然后再采用渐近线辐射传输模型(ART)计算冰雪的BRDF和反照率。对于海冰/海水的混合像元，采用蒙特卡洛( Monte-Carlo)的方法随机生成像元中的海冰覆盖比例，然后把冰雪和海水BRDF进行线性混合构建得到海冰的BRDF和反照率。基于上述模型，设置典型参数进行模拟，建立冰雪、海水和冰水混合三种地表的BRDF和反照率测试数据集。然后采用直接反演算法思路和角度网格划分策略，建立了反演这三种地表的反照率的角度网格回归算法。该算法具有简单、针对性强和性能稳定的优点，已经在我国自主生产的全球陆表参数产品（GLASS）中得到应用。本研究还针对GLASS主产品分辨率较低（1km）的特点开发了250m反照率产品算法。. 本研究丰富了定量遥感模型与方法，为我国生产完全覆盖地球表面的BRDF/反照率产品奠定算法基础。基于海洋反照率模型以及生产的冰雪和海冰反照率产品开展了反照率与气候变化关系的分析。并在国内外学术杂志上发表了相关研究成果。