Synergy of AERONET and MODIS AOD products in the estimation of PM2.5 concentrations in Beijing

Disong Fu,Xiangao Xia,Xiaojing Li,Jianzhong Liu,Xiaoling Zhang,Jun Wang

doi:10.1038/s41598-018-28535-2

Abstract

Satellite aerosol optical depth (AOD) is widely used to estimate particulate matter with aerodynamic diameter ≤2.5 µm (PM2.5) mass concentrations. Polar orbiting satellite retrieval 1–2 times each day is frequently affected by cloud, snow cover or misclassification of heavy pollution. Novel methods are therefore required to improve AOD sampling. Sunphotometer provides much more AODs than satellite at a fixed point. Furthermore, much of the aerosol pollution is regional. Both factors indicate that sunphotometer has great potential for PM2.5 concentration estimation. The spatial representativeness of the Aerosol Robotic Network (AERONET) AOD at Beijing site is investigated by linear regression analysis of 13-year daily paired AODs at each grid from Moderate Resolution Imaging Spectroradiometer (MODIS) on Aqua and Beijing AERONET. The result suggests a good correlation for the whole Beijing Administrative region, with regional mean correlation coefficient exceeding 0.73. Pixel AODs are then estimated from AERONET AOD using linear equations, which are verified to have the same accuracy as that of MODIS AOD. Either AOD from MODIS retrieval or estimation from AERONET AOD in the absence of MODIS pixel AOD is finally used to predict PM2.5 concentration. Daily AOD sampling in average is enhanced by 59% in winter when MODIS AODs are very limited. More importantly, synergy of AERONET and MODIS AOD is able to improve the estimation of regional mean PM2.5 concentrations, which indicates this method would play a significant role in monitoring regional aerosol pollution.

Highlights

These studies differ to some extent in their methodologies, the fundamental requirement of these methods is the same, i.e., satellite aerosol optical depth (AOD) products should be available; otherwise, it is all but impossible to derive PM2.5 from AOD with sufficient observational constraint
The boundaries of the region resembling AOD temporal variability as that at Aerosol Robotic Network (AERONET) sites vary between 200 and 500 km depending on their specific locations, which indicates temporal variation of AOD at AERONET sites could be representative for a larger region[13,14,15]
Validation shows that this method is robust in the Beijing Administrative area that suggests a great potential of AERONET AOD products for monitoring PM2.5 concentration, especially in heavily polluted regions

Summary

Result

MODIS retrieves AOD at approximately 50% of probability (regional mean). Regional mean AOD sampling over entire area increases to 81%. MODIS and fused AODs were compared with independent sunphotometer measurements at Xianghe and SDZ (Fig. 4). Fusion of AERONET-derived pixel AOD and MODIS AOD results in an increase of AOD sampling by 65% at Xianghe and 93% at SDZ in the winter (Fig. 4), which definitely would be expected to improve PM2.5 estimation from AOD

XiangHe

Findings

Discussion and Conclusions