Adaption of the MODIS aerosol retrieval algorithm by airborne spectral surface reflectance measurements over urban areas: a case study

Abstract

Abstract. MODIS retrievals of the aerosol optical depth (AOD) are biased over urban areas, where surface reflectance is not well characterized. Since the operational MODIS aerosol retrieval for dark targets assumes fixed spectral slopes to calculate the surface reflectance at 0.47 μm, the algorithm may fail in urban areas with different spectral characteristics of the surface reflectance. To investigate this bias we have implemented variable spectral slopes into the operational MODIS aerosol algorithms of Collection 5 (C5) and C6. The variation of slopes is based on airborne measurements of surface reflectances over the city of Zhongshan, China. AOD retrieval results of the operational and the modified algorithms were compared for a MODIS measurement over Zhongshan. For this case slightly lower AOD values were derived using the modified algorithm. The retrieval methods were additionally applied to MODIS data of the Beijing area for a period between 2010–2014 when also AERONET data were available. A reduction of the differences between the AOD retrieved using the modified C5 algorithm and AERONET was found, whereby the mean difference from 0.31 ± 0.11 for the operational C5 and 0.18 ± 0.12 for the operational C6 where reduced to a mean difference of 0.09 ± 0.18 by using the modified C5 retrieval. Furthermore, the sensitivity of the MODIS AOD retrieval for several surface types was investigated. Radiative transfer simulations were performed to model reflectances at top of atmosphere for predefined aerosol properties. The reflectances were used as input for the retrieval methods. It is shown that the operational MODIS AOD retrieval over land reproduces the AOD reference input of 0.85 for dark surface types [retrieved AOD = 0.87 (C5)]. An overestimation of AOD = 0.99 is found for urban surfaces, whereby the modified C5 algorithm shows a good performance with a retrieved value of AOD = 0.86.

An Evaluation of Four MODIS Collection 6 Aerosol Products in a Humid Subtropical Region

“…However, both the DT 10 km and 3 km products exhibit a positive bias at the "Zhongshan Univ" site. This is because the DT algorithm typically underestimates surface contributions over highly urbanized areas with limited vegetation cover, resulting in the overestimation of AOD values [35]. Although the number of MODIS/AERONET collocations at the "Zhongshan Univ" site is small, the results demonstrate the limitations of the DT algorithm over non-vegetated urban areas, which is consistent with previous reports [15,18].…”