Interannual Variability of the Wintertime Polar Vortex in the Northern Hemisphere Middle Stratosphere1

Transcription

1 February 1982 j. M. Wallace and Fong-Chiau Chang 149 Interannual Variability of the Wintertime Polar Vortex in the Northern Hemisphere Middle Stratosphere1 By John M. Wallace and Fong-Chiau Chang Department of Atmospheric Sciences, University of Washington (Manuscript received 2 October 1981, in revised form 9 November 1981) Abstract Interannual vartability of the Northern Hemisphere polar vortex at the 30mb during 21 winter seasons is examined making use of two indices which may be viewed as measures of the intensity of the vortex. The signature of stratospheric warmings is clearly evident in time series of the indices, but these events account for only part of the interannual variability. Alleged relationships between the intensity of the wintertime stratospheric polar vortex and the phase of the equatorial stratospheric quasi-biennial oscillation and the tropical tropospheric Southern Oscillation are examined, making use of the same indices. Both relationships show up quite distinctly in our analysis but neither one is consistent with the anomalies in the intensity of the polar vortex during all winters. 1. Introduction Recent studies by Holton and Tan (1980) and van Loon et al. (1981) have rekindled interest in the interannual variability of the high latitude wintertime stratospheric circulation and its possible relationship to phenomena in the tropics: the former paper was concerned with the quasibiennial oscillation of zonal wind in the equatorial stratosphere, and the latter with the Southern Oscillation. For a further discussion of these studies, the reader is referred to an article by Holton and Tan in this issue. In the above investigations it was necessary to contrast the gross features of the stratospheric circulation between groups of individual winters. If one were asked to perform such an analysis on the tropospheric circulation, one would need some a priori specification of the features that were to be compared. Otherwise it would be extremely difficult to obtain statistically significant results, since there are so many different ways in which one winter's circulation may be different from another's. Fortunately, the stratospheric circulation has considerably fewer degrees 1 Contribution number 518, Department of Atmospheric Sciences AK-40 University of Washington, Seattle, Wash of freedom; most of the interannual variability is concentrated in the zonally symmetric component, which has a rather simple meridional structure and is directly related to the intensity of the polar vortex [e.g., see van Loon et al., (1975.) and the studies mentioned above]. However, the interpretation of the winter to winter differences is complicated by the occurrence of midwinter stratospheric warmings which are capable of producing very large circulation changes on time scales of a few weeks. It would be useful to know how much of the interannual variability is associated with differences in the intensity and timing of stratospheric warmings and how much of it is a reflection of differences in the strength of the polar night vortex during relatively undisturbed periods. Holton and Tan made some attempt to address this question by separating the winters into early (November- December) and later (January-March) parts in the belief that the major effects of warmings would be felt in the later part of the winter. The objectives of the present study are: (i) to demonstrate that it is possible to characterize the intensity of the wintertime stratospheric polar vortex by means of a single index. (Specifically, we will show that two rather different circulation indices exhibit similar time dependence.)

2 150 Journal of the Meteorological Society of Japan Vol. 60, No. 1 (ii) to use two such indices to document the temporal variability of the stratospheric circulation during the winters through on a monthly basis and, during most of the period, on a 10 day basis. (iii) to use the documentation in (ii) to assess the recent results of Holton and Tan (1980) and van Loon et a. (1981) concerning possible effects of the Quasi-Biennial and Southern Oscillations on the extratropical wintertime stratospheric circulation. 2. Data This study is based on hemispheric, gridded 30mb height data obtained on magnetic tape from the Data Library of the National Center for Atmospheric Research. From the January 1962 winter onward the data are derived from the operational, objectively analyzed products prepared by the United States National Meteorological Center. These data have been processed in terms of 10 day averages (Days 1-10, 11-20, and 21 onward, for each calendar month). Monthly mean data for the winters November 1957 through December 1961 are based on hand drawn analyses made at the Free University of Berlin as published regularly in Meteorologische Abhandlungen. These analyses must be viewed with caution, since data from the U.S.S.R. did not become routinely available until April (K. Labitzke2). 3, Circulation indices We define the following two measures of the intensity of the winter-time polar vortex at the 30mb level: C1: the difference between the zonally averaged 30mb geopotential height on the 40*N and 70*N latitude circles, and C2: the minimum 30mb height anywhere in the hemisphere, based upon the 10 day or monthly mean map for the period in question. The first index is a measure of the zonally averaged, geostrophic zonal wind component in a broad latitude belt which encompasses the strongest westerly winds associated with the polar vortex. The second is a much more local measure of the intensity of the polar vortex, irrespective of its position relative to the North Pole or its meridional extent. If the time variability of the stratospheric circulation is dominated by the zonally symmetric component which has a simple and well defined latitudinal structure, C1 and C2 should show a strong negative temporal correlation. On the other hand, if temporal variability of the planetary waves is of primary importance, then C1 and C2 might be expected to behave relatively more independently of one another. For example, C1 should be much more sensitive to the amplitude of the planetary waves than C2, and C2 should be sensitive to the relative phases of zonal wavenumbers 1 and 2 at high latitudes whereas C1 should not be. In order to show the time variations in the indices, we have presented, in Tables 2 and 3, time series of anomalies based on departures from the climatological mean values shown in Table 1. Since numerical values of C2 are Table 1: Climatological mean values of C1 (the difference between the zonally averaged 30mb geopotential height on the 40N and 70N latitude circles, in units of decameters) and C2 (the minimum 30mb height anywhere in the hemisphere, based on the 10 day or monthly mean map for the period in question, in units of decameters). Means are computed separately for the period indicated. (a) Winters through (b) Winters through Personal communication.

3 February 1982 j. M. Wallace and Fong-Chiau Chang 151 Table 2: Anomalies in Index Cl from Climatological means listed in Table 1 for periods as indicated. Table 3: Anomalies in Index C2 from climatological means listed in Table 1 for periods as indicated.

4 152 Journal of the Meteorological Society of Japan Vol. 60, No. 1 dependent upon the length of the averaging interval, separate climatologies have been prepared for the monthly mean and 10-day. mean data; and the anomalies described in subsequent sections represent departures from those two separate climatologies. Hence in Tables 2 and 3 the top five entries in any column should sum to zero and the entries in any column for the sixteen years of 10-day mean data should also sum to zero. The use of separate climatologies introduces a certain amount of inhomogeneity into any statistics based on the full 21-year data set, but we don't believe that this should pose any serious problems for the purposes of this study. Note that positive anomalies in Cl indicate a strong polar vortex, while positive anomalies in C2 are indicative of a weak polar vortex. 4. Time variations in the indices It is evident from Tables 2 and 3 that major midwinter breakdowns of the polar night jet are marked by large changes in both indices (decreases in Cl and increases in C2). Well known examples include the late January warming of and the event that took place around the first of January, There are abrupt changes in C 1 and C2 in the same sense in January, 1960, January 1968 and February However it is evident that not all the large anomalies in Cl and C2 are associated with such "stratospheric warming signatures", or the lack thereof. Examples of anomalies not associated with sudden breakdowns of the polar vortex include the and winters, which were characterized by a persistently strong polar vortex, and the winters of and , which were characterized by a weak vortex which set in gradually during December and persisted until late winter. There is significant interannual variability during November and December which is not associated with major breakdowns of the polar vortex. Strong midwinter warmings can occur in winters in which the polar vortex averages out to be stronger than normal (e.g., Table 4: Anomalies from the 21 year means of the Indices Cl, C2, and C3 for winter months as indicated. C3 is defined near the end of Section 4 in the text. ND refers to a November-December average, DJF to December through February, etc. C3 applies to winter seasons as a whole.

5 February 1982 J. M. Wal lace and Fong-Chiau Chang ). Hence there does not appear to be any simple way to characterize the time variability of the intensity of the wintertime stratospheric polar vortex. Sudden warmings make a strong contribution to this variability, but they are not the only factor. Table 4 shows a summary of the interannual variability of C1 and C2, averaged for November and December (ND), and December through February (DJF), January through March (JFM) and November through March (N-M). The results in this Table have been obtained by averaging the data in Tables 2 and 3 over the appropriate time intervals. The temporal variability described in this section is common to both indices C1 and C2. Therefore it can be assumed to be representative of broad scale changes in the zonally symmetric component of the stratospheric circulation, similar to those reported by Holton and Tan (1980) and van Loon et al. (1981). As expected the two indices show a strong negative temporal correlation, particularly in the time averaged results, as indicated in Fig. 1. The index C3 in the last column of Table 4 has been formed by taking the difference C1-C2 from the midwinter months (DJF) and for the entire winter (N-M) and adding them together. It may be regarded as an indicator of the strength of the polar vortex over the whole winter, with emphasis on the midwinter months. Positive values for a given winter indicate that the polar vortex is stronger than normal. 5. Relation to quasi-biennial oscillation In Table 5 the twenty-one winters included in this study have been ranked in accordance with the value of C3. The winters with a strong polar vortex at the 30mb level are near the top of the list, and those with a weak vortex are near the bottom. Obviously, other criteria could have been chosen as a basis for the ranking, but any reasonable combination of C1 and C2 averaged over the winter would yield similar results in the sense that winters near the top or bottom of the list in Table 5 would tend to remain there. The third column gives Holton and Tan's classification in terms of easterly (E) or westerly (W) categories based on values of the November-December zonally averaged zonal wind at the 50mb level in the equatorial belt, as determined from the time-height section presented by Coy (1979). Numerical values of the zonal wind, as listed in Table 1 of Holton and Tan (1981) are given in the fourth column of the Table. The tendency for winters in the westerly phase Table 5: Winter seasons ranked in accordance with the value of C3 in Table 4, shown with classifications and statistics relating to the equatorial stratospheric quasi-biennial oscillation and the tropical tropospheric Southern Oscillation. See Sections 5 and 6 of the text for further explanation. Fig. 1 Plot of anomalies in Cl vs. those in C2. Dots represent individual 10-day averages listed in Tables 2 and 3. Circles represent December- February averages listed in Table 4. a

6 154 Journal of the Meteorological Society of Japan Vol. 60, No. 1 of the quasi-biennial oscillation (QBO) to be that the same sea-surface temperature fluctuations characterized by a strong polar vortex and vice in the equatorial Pacific are accompanied by versa, was noted previously by Holton and Tan simultaneous fluctuations in 200mb height (1980). Such behaviour is also evident in Table throughout the tropical belt (25N-25S). Their 5, where westerly winters tend to lie near the top normalized index of tropical 200mb height. as of the list and easterly winters near the bottom. computed for the months of December, January The case in favor of the existence of such a and February is listed in the last column of relationship would be quite compelling, were it Table 5. not for the existence of an unusually weak vortex In agreement with the results of van Loon et during the, winter, which corresponded al., the correspondence between HD winters and to the westerly phase of the QBO. There is no winters with a strong polar vortex and vice versa obvious way of rationalizing the seemingly contradictory result for this particular winter. From exceptions here also; most notably the winters is quite impressive. However, there are some Tables 2 and 3 it is evident that the polar vortex of and which were distinctively "low wet" winters with a relatively strong polar was weaker than normal from mid-december onward. Inspection of Coy's time-height section3 vortex. The 200mb index also shows some does not reveal any striking differences between evidence of a relationship to C3; winters with and the other winters in the Holton and low tropical 200mb heights (and a presumably Tan "westerly" category. Less striking but still weaker than normal tropospheric jetstream) tend important discrepancies are found in the winters to have a stronger than normal polar vortex. of , , and The phase Holton and Tan (1982) have also noted an of the QBO during winter is not clear, inverse relationship between the strength of the since the transition form the westerly phase to tropospheric jetstream and the polar night jet in the QBO to the easterly occurred during December Despite the exceptions noted above, connection with the QBO. Holton and Tan have demonstrated that the 7. Discussion relationship between the strength of the wintertime polar vortex and the QBO is strong enough impossible to make any definitive judgment as On the basis of the results presented here it is to easily satisfy the conventional criteria for to whether the strength of the polar vortex in statistical significance, provided that one accepts the wintertime stratosphere is related to either the assumption, implicit in such tests, that there the Quasi-Biennial Oscillation, the Southern is some a priori basis for believing that such a Oscillation or to both. Within the limited period relationship should exist. of record under consideration, it has happened (quite by coincidence, we believe) that most 6. Relationship to the Southern Oscillation winters which coincided with the easterly phase The fifth column of Table 5 shows the classification of the winters in terms of precipitation equatorial Pacific (the "low wet" condition of of the QBO were characterized by a warm in the equatorial Pacific and sea-level pressure van Loon et al.), and vice versa. It is apparent at a number of tropical stations as determined from Table 5 that the only two winters in which by van Loon et al. (1981). Their "high dry" the phases of the two oscillations are clearly (HD) winters are characterized by high pressure juxtaposed in the opposite manner are the winters in the South Pacific relative to Indonesia and of and It is to these exceptional tropical Australia and below normal precipitation in the equatorial central Pacific. Such which of the apparent relationships was more winters that we looked with hopes of determining winters coincide with episodes of below normal likely to be the real one. For example, if the sea-surface temperatures in the equatorial central and winters had both been characterized by a strong stratospheric polar vortex, Pacific [e.g. see Bjerknes (1969)]. Their "low wet" (LW) winters are characterized by opposite that would have argued in favor of a real relationship to the QBO. Whereas if they had conditions. Wallace and Horel (1981) showed both 3 The portion of Coy's time-height section after December was in error: a corrected version of the figure is found in the corrigendum section of J. Atwos. Sci. 37, No. 4, been characterized by a weak polar vortex, that would have argued in favor of a link with the SO. As it turns out, the polar vortex was strong during and weak during In conclusion, we are inclined to regard the

7 February 1982 J. M. Wallace and Fong-Chiau Chang 155 issue of whether the wintertime stratospheric polar vortex is influenced by the phase of the QBO or SO (or both) as unresolved for the present, and look to modeling studies and the acquisition of observational data for additional winters with hopes that more definitive evidence will be forthcoming. For an independent assessment of these relationships based upon a somewhat different analysis approach, the reader is referred to the article by K. Labitzke (1982) in this same issue. Acknowledgements We wish to thank James R. Holton for helpful discussions and John D. Horel for his help with the data processing and for supplying the values of his 200mb height index. We are also indebted to K. Labitzke for pointing out a number of mistakes in our early calculations. The work was supported by the Climate Dynamics Program in the Atmospheric Sciences Division of the National Science Foundation under Grant Number ATM References Bjerknes, J., 1969: Atmospheric teleconnections from the equatorial Pacific. Mon. Wea. Rev., 97, Coy, L., 1979: An unusually large westerly amplitude of the quasi-biennial oscillation. J. Atmos. Sci., 36, Holton, J. R., and H.-C. Tan, 1980: The influence of the equatorial quasi-biennial oscillation on the global circulation at 50mb. J. Atmos. Sci., 37, , - and -, 1982: The quasi-biennial oscillation in the Northern Hemisphere lower stratosphere. J. Meteor. Soc. Japan, 60, to be published. Horel, J. D., and J. M. Wallace, 1981: Planetary scale atmospheric phenomenon associated with the Southern Oscillation. Mon. Wea. Rev., 109, Labitzke, K., 1982: On the interannual variability of the middle stratosphere during the northern winters. J. Met. Sac. Japan, 60, to be published. van Loon, H., R. A. Madden, and R. L. Jenne, 1975: Oscillation in the winter stratosphere, Part I. Mon. Wea. Rev., 103, C. S. Zerefos and -, C. C. Repapis, 1981: Evidence of the Southern Oscillation in the stratosphere. Academy of Athens. Research Centre for Atmospheric Physics and Climatology, Publication No. 3, pp. 36. John M. Wallace and Fong-Chiau Chang Department of Atmospheric Sciences, University of Washington, U,S.A.