Kevin Hamilton, International Pacific Research Center, University of Hawaii, Honolulu, Hawaii, USA
George J. Boer, Canadian Centre for Climate Modelling and Analysis, Meteorological Service of Canada, Victoria, British Columbia, Canada

The quasi-biennial oscillation (QBO) of the prevailing zonal wind in the tropical stratosphere is probably the most predictable aspect of the circulation anywhere in the atmosphere (other than the astronomically-forced annual and diurnal cycles, of course). The zonal-mean zonal winds in the tropical stratosphere can be quite skillfully forecast for at least several months in advance. If the stratospheric QBO systematically (and significantly) affects the tropospheric circulation, then the QBO potentially provides some useful contribution to seasonal tropospheric forecasting. Unfortunately, the stratospheric QBO itself is not likely to be represented well in current numerical models used for seasonal predictions, and thus any potential that the QBO offers for seasonal forecasting is probably not exploited well by current numerical-model based forecast systems.

We propose to investigate how the slowly-changing QBO component of the circulation may be better exploited in seasonal prediction. We focus on the effects of better representation of the QBO in seasonal forecasts for the Northern Hemisphere winter extratropical tropospheric circulation. We will use the SMIP data to examine how current numerical model seasonal ensemble forecasts can be improved by adding a statistical correction based on the phase of the QBO.

We will first examine the boreal winter seasonal forecasts from the various SMIP models to see how much of the observed stratospheric QBO-troposphere signal appears in the forecasts. We expect to see little sign of systematic QBO dependence of the NH winter tropospheric circulation in the model forecasts. We will then add statistical corrections based on the phase of the QBO, initially basing the QBO phase used as input simply on the zonal wind near the equator at one level at the beginning of the forecast. We will also pursue two refinements to this procedure. First we will consider definitions of the phase of the QBO that combine information from the wind values over the full depth of the equatorial lower stratosphere. A second refinement will be using as input into the statistical correction a forecast QBO phase (say for the midpoint of the forecast period) rather than just the initial phase.