Sensitivity of the Arctic atmospheric boundary layer process to sea ice concentration: Polar WRF forced with multiple sea ice datasets

Sea ice concentration plays a fundamental role in the Arctic atmospheric boundary layer (ABL) process by modulating the exchange of water, momentum and heat. While there are a number of satellite-based sea ice concentration datasets which all agree each other in terms of long-term variability and trend, these datasets also exhibit significant variations on synoptic time-scales due to different algorithms. This study attempts to quantify the sensitivity in synoptic-scale ABL and surface energy balance to such uncertainty by making use of the Polar WRF model forced with three widely used sea ice concentration datasets. The 15-month-long Polar WRF simulations forced with daily sea ice data from 1) NASA Team algorithm (NT), 2) NASA Bootstrap algorithm (BT) and 3) EUMETSAT OSI SAF Global Sea Ice Concentration Reprocessing Data are compared with the available in situ data from the North Pole drifting station over the pack ice, SHEBA Ice Station over the multi-year ice, and the R/V Mirai near the sea ice margin in Beaufort Sea.

The ABL and flux fields from the Polar WRF simulations agree well with those from the in situ measurements, although large mean bias is found in the cloud and radiative flux fields. NT with the least sea ice concentration features more unstable ABL and hence higher 10-meter wind compared to BT with the highest sea ice condition. Near-surface geostrophic wind based on sea level pressure, in contrast, show a minimal response to the sea ice conditions, suggesting the important linkage between sea ice condition and local turbulent boundary layer wind, rather than syncopic geostrophic wind, through stability adjustment. There is a quasi-linear relationship in all ABL and flux fields examined with the sea ice concentrations. The slope of the linear fit, representing the degree of sensitivity to uncertainty in sea ice concentration, is largest during the winter when the uncertainty in sea ice concentration is small, while the lowest sensitivity is found when the uncertainty is most pronounced. More detailed across-model sensitivity to sea ice concentration is discussed. Overall, the results suggest that the substantial sensitivity in Arctic ABL is associated with the uncertainty in the satellite-based estimates of the sea ice concentration, which may translate into the uncertainty in model results and predictions from the Arctic Ocean modeling.