Understanding the role of the extratropical stratospheric circulation in subseasonal prediction of surface temperature within a multiple linear regression framework

Many of the key sources of predictability of North American climate on subseasonal to seasonal (S2S) timescales originate in the tropics. While these sources, including the Madden Julian Oscillation (MJO) and El Niño Southern Oscillation (ENSO), have helped to increase the skill of S2S forecasts, extratropical sources of predictability could potentially further enhance forecast skill particularly when ENSO/MJO are in neutral/inactive phases. This study examines the extent to which the extratropical stratospheric circulation can be leveraged to improve North American temperature forecasts on S2S timescales. In particular, we incorporate stratospheric-based predictors, each characterizing different aspects of extratropical stratospheric variability, into a multiple linear regression (MLR) tool currently in operations at the Climate Prediction Center. This MLR tool, which currently predicts temperature and precipitation at Weeks 3-4 using MJO, ENSO, and a linear trend, is extended by adding each stratospheric predictor separately. Initial evaluation of the cross-validated skill of the expanded MLR demonstrates improvement in temperature forecasts relative to the original MLR, particularly during boreal autumn and winter. Of all the stratospheric-based predictors tested, indices that characterize stratosphere-troposphere coupling provide the greatest improvements in temperature forecast skill during some sudden stratospheric warmings but also during other stratospheric polar vortex events.