What is the lead-time for a skillful winter forecast?

The Arctic Oscillation (AO) is the dominant mode of Northern Hemisphere (NH) winter climate variability. Most scientists consider the AO to be a result of weather noise and therefore unpredictable beyond synoptic timescales. We will show that the winter AO can be predicted with a lead-time of up to two months, however there is a well-defined barrier to that predictability. One universal feature of the AO pattern is the inverse relationship between climate anomalies in the high and mid-latitudes. For example during a negative AO phase, pressures are high and temperatures are warm at high latitudes while pressures are low and temperatures are cold in the mid-latitudes. Therefore using a polar cap average of atmospheric variables as a proxy of the AO, we will show that lower tropospheric circulation anomalies and boundary conditions from the fall can be used as leading indicators for the winter circulation anomalies. We will also show using the same diagnostics that the limit of predictability is October, with possibly some highly degraded predictability extending back into September. This predictability may or may not be independent of ENSO. However even if the phase and amplitude of the winter ENSO can be successfully predicted earlier than October, ENSO's demonstrable impact on winter weather in the mid-latitudes is regionally limited to only a fraction of the NH landmasses.