92nd American Meteorological Society Annual Meeting (January 22-26, 2012)

Tuesday, 24 January 2012: 3:30 PM
A Framework for Incorporating MJO and ENSO Information Into Probabilistic Extended Range Forecasts
Room 354 (New Orleans Convention Center )
Nat Johnson, Univ. of Hawaii, Honolulu, HI; and E. E. Riddle, S. B. Feldstein, M. L'Heureux, D. C. Collins, M. Stoner, and S. P. Xie

One challenge for forecasters is finding a way to incorporate known relationships among the Madden-Julian Oscillation (MJO), El Niņo-Southern Oscillation (ENSO), and the large-scale atmospheric circulation into existing extended-range forecast products for the one-to-four week time period. In an effort to provide a framework for incorporating these influences, we present results that demonstrate a strong influence of the MJO, ENSO, and two new climate indices, a Rossby wave source and storm track index, on the frequency of geopotential height patterns over the extended Pacific/North America region. With the use of k-means cluster analysis to approximate the continuum of wintertime 500 hPa geopotential height patterns over this region, we demonstrate that the combined influence of the MJO and ENSO produces statistically significant shifts in the cluster pattern frequency distribution for time lags of one to four weeks. In particular, we identify three clusters of 500 mb height anomalies with frequencies that are particularly influenced by the MJO. The first resembles the negative phase of the Arctic Oscillation (AO)/North Atlantic Oscillation (NAO). The second (third) corresponds with a positive (negative) phase of the Pacific/North American pattern (PNA) together with a weakly negative (positive) AO. The MJO influence, particularly in the PNA-dominated clusters, is strongly modulated by ENSO. The frequency shifts we observe in these regimes are generally consistent with results from previous studies, but here we demonstrate that these statistically significant frequency changes extend to longer lead times of up to 30 days. We then incorporate these relationships into a Bayesian framework that combines this information with Climate Forecast System Version 2 (CFSv2) forecast performance based on archived hindcasts for the purpose of generating extended-range forecasts in the one-to-four week time period.

Supplementary URL: