Publications

Song, Z; Liang, SL; Wang, DD; Zhou, Y; Jia, AL (2018). Long-term record of top-of-atmosphere albedo over land generated from AVHRR data. REMOTE SENSING OF ENVIRONMENT, 211, 71-88.

Abstract
Top-of-atmosphere (TOA) albedo is a fundamental component of Earth's energy budget. To date, long-term global land TOA albedo products with spatial resolution higher than 20-km are not available. In this study, we propose a novel algorithm to retrieve TOA albedo from multispectral imager observations acquired by Advanced Very High Resolution Radiometer (AVHRR), which provides the longest continuous record of global satellite observations since 1981. Direct estimation models were established first to derive instantaneous TOA broadband albedo under various atmospheric and surface conditions, including cloudy-sky, clear-sky (snow-free) and snow cover conditions. To perform long-term series analysis, the instantaneous TOA albedo were then converted to daily/monthly mean values based on the diurnal curves from multi-year Clouds and the Earth's Radiant Energy System (CERES) 3-hourly flux dataset. Calibration differences between sequential AVHRR sensors were further mitigated by invariant targets normalization. The retrieved TOA albedo at 0.05 degrees x 0.05 degrees was validated against two TOA albedo datasets, CM SAF (Climate Monitoring Satellite Application Facility) flux data and CERES flux data, at spatial resolutions of 0.05 degrees x 0.05 degrees, 20 km x 20 km and 1 degrees x 1 degrees. The instantaneous TOA albedo had an overall Root-Mean-Square-Error (RMSE) of 0.047 when compared with 20-km CERES fluxes, whereas the 1 degrees by 1 degrees monthly mean TOA albedo showed closer agreements with both CM SAF and CERES, with RMSE ranging from 0.029 to 0.040 and from 0.022 to 0.031, respectively. Moreover, our product was found to be highly consistent with both CERES and CM SAF at long-term trend detection. The extensive validation indicated the robustness of our algorithm and subsequently, comparable data quality with existing datasets. With global coverage and long time series (1981-2017), our product is expected to provide valuable information for climate change studies.

DOI:
10.1016/j.rse.2018.03.044

ISSN:
0034-4257