Publications

Gao, L; Li, J; Chen, L; Zhang, LY; Heidinger, AK (2016). Retrieval and Validation of Atmospheric Aerosol Optical Depth From AVHRR Over China. IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 54(11), 6280-6291.

Abstract
As Advanced Very High Resolution Radiometer (AVHRR) lacks a 2.1-mu m band, the widely used "dark target" algorithm cannot be used to retrieve aerosol optical depth (AOD) from AVHRR over land. Instead, a multiple regression algorithm has been developed to process a time series of AVHRR Level_1b measurements over China (15 degrees-45 degrees N, 75 degrees-135 degrees E) for AOD retrieval. As the apparent reflectance of AVHRR is closely related to AOD, which can be provided by Moderate Resolution Imaging Spectroradiometer (MODIS), spatially and temporally collocated Aqua/MODIS AOD and AVHRR Level_1b measurements from four years (January 2008-December 2011) were used to generate the regression coefficients. Angle information, normalized difference vegetation index, water vapor, and surface elevation are chosen in addition to the apparent reflectance as predictors for different surface types. By applying the regression coefficients to AVHRR, the AOD product from independent AVHRR Level_1b measurements (May 2003-December 2007) was generated. Validation with AErosol RObotic NETwork (AERONET) AOD and comparison with MODIS AOD products have been conducted to evaluate the uncertainty of the AVHRR AOD from May 2003 to December 2007. The distribution pattern of the seasonal mean AOD from AVHRR is consistent with that of the MODIS AOD. Taking regions with rapid economic development, for example, the regional monthly mean AOD for these two data sets agrees well, with a consistent tendency and high correlation coefficients. When compared with AERONET from four sites in China, the results are also encouraging. Results show that the multiple regression method offers the potential to generate an AOD climatology data record from a long-term AVHRR Level_1b data set over land.

DOI:
10.1109/TGRS.2016.2574756

ISSN:
0196-2892