Forecasting North Pacific Height Anomalies with the MJO on S2S timescales

North Pacific blocking has been proven difficult to predict beyond 10 days lead time in the NCEP Global Ensemble Forecast System (GEFS) and the discrepancy between model prediction skill and estimates of its predictability implies room for improvement (Hamill and Kiladis, 2014). In addition, Henderson et al. (2015) showed that the frequency of North Pacific blocking is a function of the Madden Julian Oscillation (MJO) phases based on RMM indices (Wheeler and Hendon, 2004) as well as lead time, which also implies that the potential exists for extended prediction of blocking events. Specifically, their work demonstrated a dramatic decrease (increase) in blocking episodes over the North Pacific during MJO phases 2 and 3 (phases 7 and 8).

In our study, we explore the relationship between the MJO and 500 hPa geopotential height (Z500) over the North Pacific at S2S timescales (2 weeks - 4 weeks) using the MERRA reanalysis and compare the results to ECMWF ensemble hindcasts. We find that some MJO phases are followed by robust tropical-extratropical teleconnections (phases 2, 3, 6, 7 and 8) while others are not (phases 1, 4 and 5). Results indicate that when the tropical-extratropical teleconnections are more robust, the variance of Z500 decreases dramatically. This signal of low variance evolves with MJO phase and lead time. Furthermore, the robust teleconnections exhibit higher agreement on the sign and amplitude of Z500 across the 10 ensemble members in the ECMWF hindcasts extending the models’ prediction skill of Z500 by 6 days.