北京大气PM2.5遥感监测业务化方法探讨

为探索卫星遥感监测大气ρ(PM_2.5)业务化方法,以北京为例,利用2013年MODIS卫星资料和北京35个地面自动监测站(下称自动站)的实时观测数据,以目前国内外应用最广泛的3种卫星反演大气气溶胶的方法——AOD(气溶胶光学厚度)、K_drya,0(气溶胶干消光系数)和R_a(气溶胶表观反照率)反演地面ρ(PM_2.5)的方法(分别称为AOD法、K_drya,0法和R_a法)为基础,结合地面ρ(PM_2.5)实测数据,建立了气溶胶反演参数与ρ(PM_2.5)统计关系,进一步测算了全市区域ρ(PM_2.5)的分布情况.结果表明:3种方法都具有较高的反演精度,其获取的全年ρ(PM_2.5)与地面实测数据的相关系数分别达到0.80、0.81和0.85,其中R_a法结果精度最高.从季节来看,R_a法在除夏季外的其他季节与地面监测数据相关系数都在0.70以上,优于其他2种方法.建议在春、秋、冬三季以R_a法,夏季以AOD法或K_drya,0法为基础进行北京PM_2.5业务化遥感监测.基于R_a法探讨了在2013年11月20—23日区域性大气重污染过程中北京PM_2.5区域分布特征和变化过程,卫星反演结果相对误差低于20%,直观地反映了区域大气颗粒物污染的时空分布规律.研究显示,三者都可以用来反演北京地区ρ(PM_2.5),其中R_a法最简便易行,尤其适用于业务化遥感监测. 

Routine Operational Algorithm for Remote Sensing of Atmospheric PM2.5 in Beijing

Satellite remote sensing of ground-level PM_2.5 concentration from either aerosol optical depth(AOD), surface aerosol dry extinction coefficient(K_drya,0) or aerosol albedo(R_a) has been extensively studied. To identify which one can be the best operational algorithm for estimating PM_2.5 concentration, the three aerosol-retrieval methods were compared by using data from MODIS and 35 ground-based automatic monitoring air quality sites in Beijing in 2013. The statistical relationship between the satellite-observed aerosol parameters and ground PM_2.5 was built and then applied to retrieve PM_2.5 concentrations in Beijing for each of the methods. Verification of the satellite-retrieved PM_2.5 against ground observation showed high correlation coefficients(0.80, 0.81 and 0.85 for AOD, K_drya,0 and R_a, respectively) for all three of the methods, with the highest accuracy for R_a. R_a also performed better than the other two methods in three seasons except for summer, with correlation coefficients higher than 0.70. As a result, R_a was recommended as the operational algorithm in Beijing during the spring, autumn and winter, while AOD or K_drya,0 method could be used in summer. Meanwhile, the PM_2.5 retrieved by the R_a method was used for analysis of a highly polluted event in Beijing during November 20th to 23th, 2013. Relative error of less than 20% was estimated, and the satellite-retrieved PM_2.5 well and directly displayed the spatiotemporal distribution of regional pollution. Comparison between the three methods and validation with ground observation showed that all of the three methods could be applied to retrieve PM_2.5, and R_a was the most straightforward for the operational algorithm to monitor regional air pollution.