基于多角度遥感和大气模式数据的气溶胶组份反演算法研究

The unique design of Multi-Angle Imaging Spectro-Radiometer (MISR) and its powerful retrieval algorithm can provide more particle information than many other satellite instruments. However, MISR also has limitations, such as the complex algorithm, the coarse resolution, and the limited and no prior information constrained aerosol model. First, our approach attempts to transform aerosol simulations of atmospheric chemistry model (e.g. GEOS-Chem and GOCART) to MISR mixtures, and then build the look-up table by the radiative transfer code. Second, the surface reflectance will be calculated based on the top of atmospheric reflectance of satellite and the related parameters to describe its variation in different angles. Then, the Empirical Orthogonal Function (EOF) will be generated to solve the aerosol optical depth of MISR mixtures. Third, we will conduct studies in MISR mixture disassembling, particle classification and recombination, and weight analysis, which will be applied to the calculation model for aerosol components and their optical properties. We will also jointly compare the ground-based measurements, model simulations, and satellite retrievals to evaluate their applicability and uncertainties. Finally, long-term MISR aerosol products in China will be released and used in research such as climate change, haze source apportionment and public health.

MISR独特的多角度设计和功能强大的反演算法能提供比目前其他卫星仪器更加丰富的气溶胶信息。但是MISR也存在算法难以实现、空间分辨率较粗、气溶胶模式较少且缺乏先验知识约束等缺点。本项目首先基于GEOS-Chem、GOCART等数据，将大气化学模式模拟结果转化为MISR气溶胶混合物，并通过辐射传输方程构造反演查找表。其次，借助卫星观测的表观反射率和相关大气参数模拟地表反射率，用来描述特定区域内的地表参数随波段和观测角度的变化情况，并构造经验正交函数进行求解不同气溶胶混合物下的气溶胶光学厚度。再次，开展MISR 混合物分解、组份归类与重组、权重分析等方面的研究，并将其应用到气溶胶组份和光学参数计算模型。最后，交叉对比地基观测，模式模拟，以及卫星反演结果，评价不同产品的适用性及误差来源。生产我国区域内长时间序列的气溶胶组份和光学特性产品，用于气候变化、灰霾解析、公共健康等研究领域。

多角度MISR传感器虽然能够反演高精度的气溶胶光学厚度，但由于尚未解决气溶胶混合物较少和缺乏先验知识约束等问题，使得其气溶胶微物理和成份光学参数反演方面还有一定不足。本研究基于GEOS-Chem等大气化学模式数据，解决了如何将模拟输出转换为卫星遥感反演所需的关键参量等问题，并利用地基AERONET数据对大气化学模式产品进行了精度评价。其次，本研究完成了MISR 经验正交模型构建工作，同时对比了卫星遥感和模式模拟的水溶型、沙尘型、煤烟型等成份的气溶胶光学厚度在我国的分布情况。再次，本研究利用包括地基AERONET和GOCART等先验知识对MISR的气溶胶混合物进行筛选，结果证明了新计算的气溶胶光学特性具有较大的改进。最后，本研究构建了近地面颗粒物浓度遥感估算模型并进行验证，同时生产了我国及典型地区长时间序列的气溶胶光学和颗粒成份遥感产品。