Simulation of Snow on Arctic Sea Ice Using a Coupled Snow–Ice Model

Abstract

The new Recherche Prévision Numérique (NEW-RPN) model, a coupled system including a multilayer snow thermal model (SNTHERM) and the sea ice model currently used in the Meteorological Service of Canada (MSC) operational forecasting system, was evaluated in a one-dimensional mode using meteorological observations from the Surface Heat Budget of the Arctic Ocean (SHEBA)?s Pittsburgh site in the Arctic Ocean collected during 1997/98. Two parameters simulated by NEW-RPN (i.e., snow depth and ice thickness) are compared with SHEBA?s observations and with simulations from RPN, MSC?s current coupled system (the same sea ice model and a single-layer snow model). Results show that NEW-RPN exhibits better agreement for the timing of snow depletion and for ice thickness. The profiles of snow thermal conductivity in NEW-RPN show considerable variability across the snow layers, but the mean value (0.39 W m?1 K?1) is within the range of reported observations for SHEBA. This value is larger than 0.31 W m?1 K?1, which is commonly used in single-layer snow models. Of particular interest in NEW-RPN?s simulation is the strong temperature stratification of the snowpack, which indicates that a multilayer snow model is needed in the SHEBA scenario. A sensitivity analysis indicates that snow compaction is also a crucial process for a realistic representation of the snowpack within the snow/sea ice system. NEW-RPN?s overestimation of snow depth may be related to other processes not included in the study, such as small-scale horizontal variability of snow depth and blowing snow processes.