A statistical analysis of 51 years of NCEP-NCAR reanalysis data is conducted to isolate the separate effects of the 11-yr solar cycle (SC) and the equatorial quasi-biennial oscillation ( QBO) on the Northern Hemisphere (NH) stratosphere in late winter (February-March). In a four-group [ SC maximum (SC-max) versus minimum (SC-min) and east-phase versus west-phase QBO] linear discriminant analysis, the state of the westerly phase QBO (wQBO) during SC-min emerges as a distinct least-perturbed ( and coldest) state of the stratospheric polar vortex, statistically well separated from the other perturbed states. Relative to this least-perturbed state, the SC-max and easterly QBO (eQBO) each independently provides perturbation and warming as does the combined perturbation of the SC-max-eQBO. All of these results ( except the eQBO perturbation) are significant at the 95% confidence level as confirmed by Monte Carlo tests; the eQBO perturbation is marginally significant at the 90% level. This observational result suggests a conceptual change in understanding the interaction between solar cycle and QBO influences: while previous results imply a more substantial interaction, even to the extent that the warming due to SC-max is reversed to cooling by the eQBO, results suggest that the SC-max and eQBO separately warm the polar stratosphere from the least-perturbed state. While previous authors emphasize the importance of segregating the data according to the phase of the QBO, here the same polar warming by the solar cycle is found regardless of the phase of the QBO. The polar temperature is positively correlated with the SC, with a statistically significant zonal mean warming of approximately 4.6 K in the 10-50-hPa layer in the mean and 7.2 K from peak to peak. This magnitude of the warming in winter is too large to be explainable by UV radiation alone. The evidence seems to suggest that the polar warming in NH late winter during SC-max is due to the occurrence of sudden stratospheric warmings (SSWs), as noted previously by other authors. This hypothesis is circumstantially substantiated here by the similarity between the meridional pattern and timing of the warming and cooling observed during the SC-max and the known pattern and timing of SSWs, which has the form of large warming over the pole and small cooling over the midlatitudes during mid- and late winter. The eQBO is also known to precondition the polar vortex for the onset of SSWs, and it has been pointed out by previous authors that SSWs can occur during eQBO at all stages of the solar cycle. The additional perturbation due to SC-max does not double the frequency of occurrence of SSWs induced by the eQBO. This explains why the SC-max/eQBO years are not statistically warmer than either the SC-max/wQBO or SC minimum/eQBO years. The difference between two perturbed ( warm) states ( e. g., SC-max/eQBO versus SC-min/eQBO or SC-max/eQBO versus SC-max/wQBO), is small ( about 0.3-0.4 K) and not statistically significant. It is this small difference between perturbed states, both warmer than the least-perturbed state, that in the past has been interpreted either as a reversal of SC-induced warming or as a reversal of QBO-induced warming.
