Improving Week 3–4 Temperature and Precipitation Outlooks by Incorporating the Stratospheric Quasi-Biennial Oscillation as a Predictor

Stakeholders desire accurate outlooks of temperature and precipitation at leads of 3 to 4 weeks across the United States. Such outlooks would improve the decision making process of hydrological, energy, and agricultural interests. While operational above or below normal temperature outlooks produced by NOAA’s Climate Prediction Center (CPC) are skillful overall, skillful precipitation outlooks are nearly non-existent and are still experimental. Recent research has shown that harnessing the stratospheric Quasi-biennial Oscillation (QBO) as a predictor offers promise in improving these outlooks. The QBO manifests itself in the equatorial stratosphere as a periodic 2 to 2.5 year oscillation of downward propagating easterly and westerly zonal wind regimes. Depending on its current phase, the QBO is capable of modulating tropical convection associated with the Madden-Julian Oscillation (MJO) and its teleconnection to the mid-latitudes. Thus, the QBO may offer a source of predictability beyond CPC’s empirical phase model that currently only incorporates the MJO, the El-Niño Southern Oscillation (ENSO), and the linear trend as predictors.

To test the ability of the QBO to add skill to Week 3-4 temperature and precipitation outlooks, four prediction schemes are developed using the current states of the MJO, ENSO, and QBO as predictors: an MJO only scheme, an MJO|QBO scheme, an MJO|ENSO scheme, and an MJO|ENSO|QBO scheme. Each scheme is tested in a leave-one-year-out cross-validated framework, over the 1979-2017 period, and in a real-time framework, over the 2011-2017 verification period with a 1981-2010 climate normal training period. The schemes are further tested in the real-time framework by including the trend as a predictor. In the cross-validated framework, the MJO|ENSO scheme offers the largest increase in skill to the MJO only scheme. The MJO|ENSO|QBO scheme offers additional skill, especially when forecasts are only judiciously issued when and where the combined influence of the MJO, ENSO, and QBO are observed in the training data to have modulated temperature and precipitation the greatest. In the real-time framework, skill scores are more modest when averaged over the 2011-2017 period. For precipitation, the MJO|QBO and MJO|ENSO|QBO schemes offer slightly more skill than the MJO|ENSO scheme, while the MJO|ENSO scheme is most skillful for temperature. However, individual years during the 2011-2017 period show that using the schemes including the QBO are often nearly as skillful or even more skillful than the MJO|ENSO scheme. Finally, while including the trend significantly improves temperature outlooks, it somewhat degrades precipitation outlooks. Overall, the results presented here demonstrate a modest increase in skill scores may be obtained when adding the stratospheric QBO as a predictor to Week 3-4 temperature and precipitation outlooks.