Forecasting and Reproducing Arctic Sea Ice Extent and Associated Teleconnections Using the Climate Forecast System

The NCEP Climate Forecast System (CFS) is a fully coupled atmosphere-ocean-sea ice-land surface model used for seasonal climate prediction and reanalysis. Recent studies have shown that there is evidence that recent rapid decline of autumn (SON) and winter (DJF) Arctic sea ice has an effect on mid-latitude weather patterns during the winter months. Therefore, accurate forecasts of Arctic sea ice extent (SIE) and the associated teleconnections are important. Here we evaluate 1) if the CFS can reproduce the observed variability of Arctic SIE for both the CFS reanalysis and reforecast and 2) to what extent the CFS can reproduce the observed linkages between decreasing Arctic SIE and Northern Hemispheric mid-latitude weather patterns.

Our preliminary results for the 1982-2010 period show that the CFS reanalysis captures most of the observed variability of SIE due to the assimilation of the observed sea ice, but it overestimates the SIE for both SON and DJF averages. By contrast, the 9 month 00Z CFS reforecast starting in June cannot capture year-to-year variability of the observed SIE and consistently underestimates the overall amount of SIE. In terms of the winter tropospheric atmospheric circulation linked to decreasing autumn SIE, the CFS reanalysis is able to capture the observed dipole pattern of positive anomalies over northwestern Europe and central Siberia and negative anomalies in the Mediterranean, but creates a false dipole pattern in the Pacific sector. The reforecast is unable to capture the observed pattern, except for some negative anomalies in the North Atlantic. For decreasing winter SIE, the reanalysis can capture most of the observed arc of positive anomalies stretching from northwestern Europe to eastern Siberia, but overestimates the magnitude of the anomaly. The reforecast is unable to capture this observed winter SIE pattern. In terms of winter surface temperature linked to decreasing SIE, the CFS reanalysis can capture the correct spatial patterns, but overestimates the cooling in North America and Siberia. The reforecast is unable to capture these patterns and creates broader warming over the Arctic with much less cooling and even some warming over North America and Siberia. We also note that the averaged reforecast of four ensemble members does not show improvement in the reforecast.