Retrieval of global broad band albedo from AVHRR data using high performance computing techniques

摘要

Land surface albedo is an important parameter in climate and energy balance models, since it determines the amount of solar radiation that is absorbed by the Earth to drive the surface processes. However, most climate models typically rely on albedo data sets derived from land cover maps, and remote sensing measurements could significantly improve the parameterization of the albedo. Retrieval of surface albedo from satellite measurements involves: 1. Normalizing the reflectances at the top of the atmosphere by the solar irradiance within the spectral wavelength of observation to obtain bidirectional reflectances 2. Correcting the bidirectional reflectances for atmospheric effects such as scattering and absorption. 3. Correcting the bidirectional reflectances for surface anisotropy. 4. Converting the narrow band reflectances to broad band albedo. Data from the Advanced Very High Resolution Radiometer (AVHRR) can be potentially used to determine land surface albedo, since it measures reflectances in the visible and near IR wavelengths daily over the entire globe. However previous studies of albedo retrieval from AVHRR data have been limited to coarse regional scales, since the computational and data storage requirements for processing AVHRR data at a global scale are high. In this paper, the authors demonstrate the feasibility of using high performance computing techniques to derive the global land surface albedo from AVHRR at 8 km spatial resolution.