Impact of diurnal variability and meteorological factors on the PM2.5 - AOD relationship: Implications for PM2.5 remote sensing

摘要

PM2.5 retrieval from space is still challenging due to the elusive relationship between PM2.5 and aerosol optical depth (AOD), which is further complicated by meteorological factors. In this work, we investigated the diurnal cycle of PM2.5 in China, using ground-based PM measurements obtained at 226 sites of China Atmosphere Watch Network during the period of January 2013 to December 2015. Results showed that nearly half of the sites witnessed a PM2.5 maximum in the morning, in contrast to the least frequent occurrence (5%). in the afternoon when strong solar radiation received at the surface results in rapid vertical diffusion of aerosols and thus lower mass concentration. PM2.5 tends to peak equally in the morning and evening in North China Plain (NCP) with an amplitude of nearly twice or three times that in the Pearl River Delta (PRD), whereas the morning PM2.5 peak dominates in Yangtze River Delta (YRD) with a magnitude lying between those of NCP and PRD. The gridded correlation maps reveal varying correlations around each PM2.5 site, depending on the locations and seasons. Concerning the impact of aerosol diurnal variation on the correlation, the averaging schemes of PM2.5 using 3-h, 5-h, and 24-h time windows tend to have larger R biases, compared with the scheme of 1-h time window, indicating diurnal variation of aerosols plays a significant role in the establishment of explicit correlation between PM2.5 and AOD. In addition, high cloud fraction and relative humidity tend to weaken the correlation, regardless of geographical location. Therefore, the impact of meteorology could be one of the most plausible alternatives in explaining the varying R values observed, due to its non-negligible effect on MODIS AOD retrievals. Our findings have implications for PM2.5 remote sensing, as long as the aerosol diurnal cycle, along with meteorology, are explicitly considered in the future. (C) 2016 The Authors. Published by Elsevier Ltd.