Comparison of AVHRR aerosol optical depth production against CARSNET data in China

Abstract

This study presents a dedicated evaluation of a newly released satellite aerosol product derived from Advanced Very-High-Resolution Radiometer (AVHRR) measurements by using the Deep Blue/SOAR aerosol retrieval algorithm. Aerosol product from the Chinese Aerosol Network (CARSNET) is used as the ground truth in the comparison. Comparison of AVHRR and CARSNET aerosol optical depth (AOD) products during 2006–2011 shows that the accuracy and precision of AVHRR AOD demonstrate complicated spatial and seasonal dependence. First, AVHRR AOD is in good agreement with CARSNET AOD over eastern urban areas in China. 45%–78% of data points fall within the expected error range (±0.05 ± 0.25× AOD) over there. Second, in regions with natural vegetation cover and relatively clean environment conditions, low correlation is observed due to small AOD loadings (<0.2 and < 0.6, respectively) and small AOD variations (<1.0). Third, for dust affected arid and semi-arid dry regions and western urban regions, AVHRR AOD products need further great improvement. A large regional variation of AOD bias with increasing CARSNET AOD may be due to insufficient representation of aerosol properties in the algorithm. The systematic underestimation of AVHRR AOD in most sites likely attributes to improper surface reflectance estimation, which should be furthermore improved in the deep blue algorithm. AVHRR has great potential in the retrieval of the historical AOD record from 1979 to the present, therefore, one would expect that AVHRR aerosol products would contribute to furthering our understanding of aerosol effects on climate and environment.