Advanced parameterization of broadband sea ice albedo was developed using in situ (SHEBA) data (read more).
Two methods were investigated to estimate melt pond concentrations (read more).
Broadband albedo was systematically estimated from the assimilaton of the AVHRR observations, IABP/POLES SAT, SSM/I brightness temperatures and SSM/I-derived sea ice characteristics (concentration, age, thickness) (read more).
Visible and near-infrared albedos were calculated from the AVHRR observations. Results were validated on SHEBA measurements (read more).
We compared the observed (AVHRR) and modeled (CCSM3) broadband albedo (read more).
Sensitivity analysis of the CCSM3 to uncertainties in albedo parameterization (read more).
- All of the reviewed sea ice albedo parameterizations failed to accurately reproduce the evolution of the broadband sea ice albedo during the melt season because of their crude parameterizations of the effects of snow and ice melting processes on the sea ice. Developed advanced sea ice albedo parameterization, explicitly dependent on the melt pond area fraction and depth was accurate and efficient.
- Methods, using classification techniques and linear regressions, were developed to estimate melt pond concentration on the sea ice from the AVHRR and active and passive microwave satellite observations.
- Methods, using neural network and least squares linear approximations, lookup tables, were developed to estimate visible and near-infrared albedo during melt season from the AVHRR and passive microwave observations.
- Comparative analysis of the clear-sky broadband sea ice albedo during melt season from the AVHRR observations and the high-resolution CCSM3 experiments of the 20th century climate simulation showed that modeled broadband sea ice albedo were generally overestimated, however, latitudinal, seasonal and interannual changes corresponded to the observations.
- Sensitivity experiments with the fully coupled low-resolution CCSM3 climate model have demonstrated that even relatively small error (systematic bias of 5%, SD of 2%) in summer ice visible and near-infrared albedos would cause significant deviations in simulated annual mean and summer minimum ice area and thickness. Thus, any single parameterization does not qualify to represent the summer sea ice albedo. This suggests that multi-parameterization or multi-parameter ensemble methods, where some members systematically overestimate and some underestimate sea ice albedo, may provide robust model results.