954 JOURNAL OF THE ATMOSPHERIC SCIENCES VOLUME 61 Reply

FIG. 1. EOF of the zonal-mean surface pressure in the CKV model (solid), zonal mean of the nonzonally averaged surface pressure in the CKV model (dashed), and zonal mean of the sector EOF shown in CKV’s Fig. 11a (dotted–dashed). We thank Dr. Robinson for his comments (Robinson 2004). Robinson argues that the results of Cash et al. (2002, hereafter CKV) are not inconsistent with the presence of an annular mode (AM) in the numerical model used. He suggests that the zonally asymmetric character of CKV’s AM events is due to the superposition of independent annular and nonannular features, with AMs being (presumably) a real dynamical phenomenon independent of the zonally asymmetric eddying motion. However, while it may be impossible to rule out the presence of such an AM in the CKV experiments, we do not believe that the AMs, thus interpreted, necessarily best represent the underlying dynamical behavior of the model. The AM problem has descriptive aspects and dynamical aspects. In CKV, we look for useful statistical descriptions of the spatial structure of extratropical atmospheric low-frequency variability in an aquaplanet model. We calculate the leading EOF of the hemispheric-wide surface pressure and the leading EOF of the zonal-mean surface pressure. We find that these two structures have practically identical meridional structures (Fig. 1); that is, the zonal mean of the hemisphericwide surface pressure EOF is the same as the EOF of the zonal-mean surface pressure. The meridional structure of the AMs is similar to the zonal mean of circulation anomalies in smaller zonal sectors. One way of seeing this is to calculate the leading EOF of the surface pressure in a 908 longitude sector (CKV, Fig. 11). The zonal mean of this structure, shown as a dotted line in Fig. 1, is also very similar to the other two curves in the figure. Another way to obtain maps of characteristic localized circulation anomalies is