大型流行病学暴露研究要求在大范围内进行长期空气质量的监测，然而我国大部分地区的PM2.5地面监测网却是近几年才开始建立的，分布不均且不具备细颗粒物组份的监测能力。本项目首先基于地基观测和大气化学模式研究结果，构建适合我国污染特征的气溶胶模式，借鉴多角度卫星反演算法思想，通过构造波段、角度组合，以及经验正交函数，实现气溶胶组份光学厚度的反演。再次，开展颗粒物估算因子的敏感性研究，探讨气溶胶廓线，气象因素，污染排放等地理信息对PM2.5组份浓度估算的影响。在此基础上，通过空间统计模型和订正模型，实现不同条件下PM2.5组份浓度的精确估算。最后，以慢阻肺等呼吸系统和心脑血管疾病为例，生产典型区域遥感产品，并对患者/死亡率等疾病数据进行时空匹配处理。通过构建细颗粒污染健康暴露统计回归模型，分析不同组份的细颗粒物浓度在不同气象暴露因素下，对公共健康的影响。

The epidemiological study requests the large-scale and long-term air quality monitoring data. However, most ground-based PM2.5 sites of China were established in megacities in recent years, and cannot provide the particulate matter components. In view of the studies on in-situ measurements and atmospheric chemistry model simulations, our approach first builds the grid-based aerosol model fitting for China’s high-pollution background. Drawing on the thoughts of Aerosol Optical Depth (AOD) retrieval using multi-angle satellite data, we will calculate the AOD for different components by using empirical orthogonal functions and a combination of wavelength and observation geometry. Second, the sensitivities of aerosol profiles, meteorological parameters, pollution emissions and other factors will be analyzed for their impact on the satellite PM2.5 estimation. With the embedded spatial structure, we develop a regression statistical model for PM2.5 prediction to better reflect the local variations of fitting parameters. Finally, taking the chronic obstructive pulmonary disease and other respiratory and cardio-cerebral vascular diseases as an example, we will produce remote sensing products on typical regions, and match them with those diseases data, such as mortality. By establishing the statistical regression model between PM2.5 and health-damaging exposure risk, the impact of polluted environment on public health will be analyzed for different constituents of particulates under different meteorological conditions.