在中国东北地区，卫星气溶胶产品尚缺乏有效的验证机制，相关应用存在较大不确定性。利用2009-2011年中国地区太阳分光光度计观测网（CSHNET）沈阳站地基观测资料，借助后向轨迹模式分析了该地区气溶胶来源和季节变化特征，并同步对比了FY-3A/MERSI和Terra/MODIS气溶胶产品精度的季节差异及误差来源。结果表明：沈阳站气溶胶光学厚度和Angstrom波长指数季节变化明显，气溶胶受远程输送和区域排放的共同影响，人为源和自然源丰富。不同粒径粒子对气溶胶光学厚度的贡献因季节而异，导致卫星反演算法中气溶胶模型选择误差较大，影响了卫星反演精度。观测期间MODIS与MERSI气溶胶产品与地基观测数据的总匹配率分别为68%和83%。当气溶胶光学厚度较低时（AOD<0.35），MODIS产品出现低估，MERSI出现高估，与实测值的相对误差分别为-46%-54%、7%-135%；当气溶胶光学厚度较高时（0.35-0.75），MODIS和MERSI均出现不同程度的低估，两者与实测值的相对误差分别为-34%-54%、-21%-75%；当气溶胶光学厚度大于0.75时，低估尤为严重，与实测值的相对误差可达-34%-100%、-9%-58%。能与地基观测匹配的MODIS和MERSI有效样本数春、秋季较高（春季分别为69%、80%，秋季分别为73%、70%），夏季次之（69%、73%），冬季最少（2%、49%）；MODIS与地基观测的相关性总体优于MERSI；但在冬季，MODIS与MERSI产品均不具备代表性。MODIS与MERSI气溶胶产品落入误差线的比例均为春季最高（22%、25%），秋季次之（19%、16%），夏季最小（6%、5%）。MERSI对粗模态气溶胶（α≤0.5）反演效果优于MODIS，而MODIS对混合模态气溶胶（0.5≤α<1.5）反演效果优于MERSI。MERSI与MODIS气溶胶产品在春、秋季可比性较好，夏季可比性较差，总体来说，春、秋季MERSI比MODIS低估更为严重。从气溶胶产品在辽宁省的区域分布来看，FY-3A/MERSI比Terra/MODIS覆盖范围略广，各季节FY-3A/MERSI与Terra/MODIS的总体空间分布特征基本一致，但对于某些地区，两者数值上依然存在较大偏差。

There still had some uncertainty when satellite aerosolproducts were applied in northeast China for lack of effective verification method. The sources and seasonal variations of aerosol were analyzed based on the HYSPLIT-4 model using the ground-based observation provided by the Chinese Sun Hazemeter Network (CSHNET) in Shenyang during 2009-2011. The quality of MERSI and MODIS Aerosol Optical Depth (AOD) products, and the source of error at four seasons were also discussed. The results showed distinctive seasonality for AOD and Angstrom exponent in northeast China. The aerosol came from remote transmission and regional emission with abundant anthropogenic and natural components. Aerosol sizes and AODs had large differences in four seasons, which leaded to a large error of aerosol model in the satellite retrieval algorithm. The accuracy of the satellite AOD products decreased due to the improper aerosol model. The comprehensive utilization ratio of MERSI and MODIS AOD products accounted for 68% and 83% of the ground-based AOD during the observing period. When AOD was lower than 0.35, the MODIS products underestimated and FY3-A products overestimated. Meanwhile, the relative error range of MODIS and FY3-A AOD products from the ground-based AOD was -46%-54% and 7%-135%. Both MODIS and FY3-A products underestimated with the relative error-34%-54% and -21%-75% when AOD was in the range from 0.35 to 0.75. A worse underestimation of MODIS and FY3-A products was found when AOD was more than 0.75 with the relative error of -34%-100% and -9%-58%. The utilization ratios of MODIS and MERSI had large differences in four seasons, 69% and 80% in spring, 73% and 70% in autumn, 69% and 73% in summer, only 2% and 49% in winter, respectively. On the whole, MODIS AOD products were more consistent with the ground-based observation than MERSI. There representativeness both of MODIS and MERSI were very poor in winter in northeast China. The percentages of MODIS and MERSI products falling within the expected errors were largest in spring (22%,25%), moderate in autumn (19%, 16%), least in summer (6%, 5%); MERSI retrieval algorithm has better capability to retrieve coarse mode aerosols (α≤0.5), while MODIS has more advantage in retrieving mixed mode aerosols (0.5≤α<1.5). The comparability between MERSI and MODIS was relatively good in spring and autumn, but poor in summer. MERSI products were more severe underestimated than MODIS in spring and autumn. Comparisons of regional distribution between MERSI and MODIS AOD products in Liaoning province showed that MERSI AOD product covered a broader range than MODIS AOD product. Although the general characteristic of AOD distribution was consistent, but there still existed obvious deviation in some areas.