卫星遥感AOD反演地面细颗粒物浓度方法与效果

The method and the correspongding effect of ground fine partical concentration retrieved by satellite remote sensing AOD

使用2015~2018年MODIS AOD产品融合地表气象资料反演了地面细颗粒物(PM2.5)浓度,并以反演的PM2.5浓度为依据,比较了地面PM2.5观测资料的各种空间插值方法.结果表明:2015~2018年反演的PM2.5平均浓度与地基观测平均浓度的R 2达0.94;干季反演效果好于湿季,珠江三角洲反演效果好于非珠江三角洲地区,原因是湿季天气系统较不稳定,非珠江三角洲地区多山脉和秸秆燃烧,导致气溶胶标高、质量消光效率等假设误差较大.使用4种插值方法对地基观测的PM2.5浓度进行插值,插值结果大致相当,反距离加权插值法较好,站点分布不均、部分区域站点密度小影响插值效果,建议在站点稀疏地区增加地面PM2.5观测站点.

Satellite remote sensing data(MODIS AOD)and ground meteorological data from 2015 to 2018 were used to retrieve PM2.5 concentration.The results showed that the retrieval was desirable,with R2 between satellite-retrieved and observed PM2.5 average concentration from 2015 to 2018 reaching 0.94.The retrieved results in dry season was better than that in dry season,results in Pearl River Delta(PRD)was better than that in Non-Pearl River Delta(N-PRD).The possible reasons were that the interpolation errors affected by assumption of aerosol scale height and mass extinction efficiency were larger both in wet seasons with more unstable weather system and in N-PRD with more mountains and straw burning.Also,four different spatial interpolation methods were compared based on in-situ PM2.5 measurement,using satellite-retrieved PM2.5 as the true values at those grids without measurement.The interpolation results were similar for these four methods,with the inverse distance weighted(IDW)interpolation method performing slightly better.Uneven site distribution and low site density in some areas might have a significant influence on interpolation performance,so we recommended deploying more ground PM2.5 observation stations in sparse areas.