Can knowledge of the state of the stratosphere be used to improve statistical forecasts of the troposphere?