Modelling and measuring the spectral bidirectional reflectance factor of snow-covered sea ice: an intercomparison study

Broadband albedo is a very important geophysical parameter in the Earth surface-atmosphere interaction in either global climate change or hydrological cycle and snowmelt runoff studies. To derive the broadband albedo accurately from satellite optical sensor observation at limited bands and at a single observation angle, the bidirectional reflectance factor (BRF) has to be specified quantitatively. In the present albedo derivation algorithms from the satellite radiance data, the BRF is either modelled or observed. Questions may arise as to how well a BRF model can be in the broadband albedo derivation. To help answer such questions, we studied the performance of a snow-surface BRF model for two specific cases under large solar zenith angles (65degrees and 85degrees). We measured snow-surface spectral directional reflectance under clear skies. The snow physical properties, such as snow grain size and snow density, at the same sites were also measured. In situ snow physical data are used to simulate the snow-surface BRF and hemispherical directional reflectance factor (HDRF) through a multilayered azimuth- and zenith-dependent plane-parallel radiative transfer model. The field measurements and BRF and HDRF simulations all reveal the forward-scattering nature of snow surface under large solar incidence angles, but the BRF model results depict the strongest forward-scattering patterns under such solar zenith angles. Because the HDRF is simulated through coupling of the surface BRF with radiative transfer in the atmosphere, the resulting HDRF patterns agree with the field measurements better than the simulated BRF does. The deviation of the simulated HDRF from field-based clear-sky directional reflectance (FCDR) is within +/- 10% for the central (viewing zenith angle <45degrees) and lateral sides of the viewing hemisphere. This level of agreement between the simulated HDRF and FCDR also implies that the simulated BRF model can provide remote-sensing estimates of spectral albedo with an uncertainty of +/- 10% for the same part of the viewing hemisphere. Further improvement in BRF model performance requires better handling of single scattering properties of snow grains, surface roughness, and atmospheric correction. Also, better procedures and techniques in field measurement are necessary for more accurate assessment of the performance of BRF models. Copyright (C) 2004 John Wiley Sons, Ltd.