Hoerling, M. P., M. Ting, and A. Kumar, 1995: Zonal flow-stationary wave relationship during El Niño: Implications for seasonal forecasting. J. Climate, 8, 1838-1852.


An analysis of the Northern Hemispheric zonal mean flow anomalies during El Niño is performed, and the dynamical effect of such atmospheric flows on the wintertime climatological stationary waves over the Pacific/North American (PNA) region is assessed. Only in the subtropical latitudes can one identify a statistically significant zonal flow anomaly during the El Niños of the historical record. Strong zonal flow anomalies in the midlatitudes are observed during individual El Niño events, although these appear to be manifestations of chaotic atmospheric behavior. The observational results are confirmed by GCM climate simulations using prescribed SSTs for the 1982-93 period. The principal SST-forced zonal flow signal in these experiments is located on the equatorward flank of the subtropical jet.

Using a linear diagnostic modes, the authors find the climatological stationary waves over the PNA region to be insensitive to zonal mean flow anomalies in the subtropics. On the other hand, zonal mean anomalies in the midiatitudes are found to induce large amplitude stationary wave anomalies, and these resemble the Pacific/North American pattern. The zonal/eddy effect may thus account for an important fraction of the interannual variability of the wintertime North American climate, although this component is evidently unpredictable from boundary-forced experiments only. Further evidence is derived from GCM hindcasts for the 1986/87 and 1991/92 El Niño winters. A lack of model skill in predicting the observed midlatitude zonal flow anomalies for these cases is shown to place a limit on the accuracy of boundary-forced simulations of North American seasonal climate anomalies.

Seasonal forecasts may yet benefit from the zonal-eddy relationship in view of the fact that the upper-tropospheric zonal flow anomalies in midlatitudes are frequently long-lived. It is thus possible that inclusion of initial atmospheric conditions, together with relevant boundary information. will be more skillful than just boundary-forced simulations alone.